Publications

Journal Articles

2025

Context-agnostic data visitation to exchange health data for scientific research: the example of PADME PHT

context-agnostic data visitation personal data exchange health research EU data protection law General Data Protection Regulation (GDPR) data sharing personal data processing health data data protection compliance individuals’ rights European Union law privacy governance research data access legal reconciliation data flow for research

Dara Hallinan , Marc Nestor , Muhammad Hamza Akhtar , Constantin Breß , Mehrshad Jaberansary , Christoph Lange , Macedo Maia , Fruzsina Molnár-Gábor , Yeliz Üçer Yediel , Zeyd Boukhers

International Data Privacy Law 2025

PDF DOI
Hallinan, D., Nestor, M., Akhtar, M. H., Breß, C., Jaberansary, M., Lange, C., Maia, M., Molnár-Gábor, F., Yediel, Y. Ü., & Boukhers, Z. (2025). Context-agnostic data visitation to exchange health data for scientific research: the example of PADME PHT. International Data Privacy Law. https://doi.org/10.1093/idpl/ipaf011

Hallinan, Dara, , and Marc Nestor, , and Muhammad Hamza Akhtar, , and Constantin Breß, , and Mehrshad Jaberansary, , and Christoph Lange, , and Macedo Maia, , and Fruzsina Molnár-Gábor, , and Yeliz Üçer Yediel, , and Zeyd Boukhers. "Context-agnostic data visitation to exchange health data for scientific research: the example of PADME PHT." International Data Privacy Law, 2025.

Hallinan, Dara, , and Marc Nestor, , and Muhammad Hamza Akhtar, , and Constantin Breß, , and Mehrshad Jaberansary, , and Christoph Lange, , and Macedo Maia, , and Fruzsina Molnár-Gábor, , and Yeliz Üçer Yediel, , and Zeyd Boukhers. "Context-agnostic data visitation to exchange health data for scientific research: the example of PADME PHT." International Data Privacy Law (2025).
@article{article11, title = {Context-agnostic data visitation to exchange health data for scientific research: the example of PADME PHT}, author = {Dara Hallinan and Marc Nestor and Muhammad Hamza Akhtar and Constantin Breß and Mehrshad Jaberansary and Christoph Lange and Macedo Maia and Fruzsina Molnár-Gábor and Yeliz Üçer Yediel and Zeyd Boukhers}, journal = {International Data Privacy Law}, year = {2025}, doi = {10.1093/idpl/ipaf011}, url = {https://doi.org/10.1093/idpl/ipaf011}, publisher = {Oxford University Press}, }
The paper examines Context-Agnostic Data Visitation (CADV) as a legal and technical mechanism for exchanging health data across institutional boundaries for scientific research. Using the PADME PHT (Personal Health Train) platform as a case study, the paper analyzes how CADV enables data access without transferring personal data, thereby maintaining compliance with the EU General Data Protection Regulation (GDPR) while still enabling distributed analytics. The paper argues that CADV represents a promising approach to reconcile the competing demands of data protection law and scientific research, particularly for health data governed by strict privacy regulations.
R4-SLAM: Toward Real-time, Robust, and Resource-Restricted Visual SLAM in Dynamic Environments

simultaneous localization and mapping (SLAM) visual SLAM dynamic environments semantic SLAM dynamic object mitigation edge devices real-time processing R4-SLAM semantic keyframe selection semantic front-end decoupling computational efficiency tracking stability processing speed accuracy evaluation TUM RGB-D dataset mobile robotics autonomous vehicles augmented reality

Zhijie Zhang , Cong Yang , Xue Qin , Wei Sui , Zeyd Boukhers , Hui Kong , Jie Liu , Zhijun Li

IEEE Transactions on Instrumentation and Measurement 2025

PDF
Zhang, Z., Yang, C., Qin, X., Sui, W., Boukhers, Z., Kong, H., Liu, J., & Li, Z. (2025). R4-SLAM: Toward Real-time, Robust, and Resource-Restricted Visual SLAM in Dynamic Environments. IEEE Transactions on Instrumentation and Measurement.

Zhang, Zhijie, , and Cong Yang, , and Xue Qin, , and Wei Sui, , and Zeyd Boukhers, , and Hui Kong, , and Jie Liu, , and Zhijun Li. "R4-SLAM: Toward Real-time, Robust, and Resource-Restricted Visual SLAM in Dynamic Environments." IEEE Transactions on Instrumentation and Measurement, 2025.

Zhang, Zhijie, , and Cong Yang, , and Xue Qin, , and Wei Sui, , and Zeyd Boukhers, , and Hui Kong, , and Jie Liu, , and Zhijun Li. "R4-SLAM: Toward Real-time, Robust, and Resource-Restricted Visual SLAM in Dynamic Environments." IEEE Transactions on Instrumentation and Measurement (2025).
@article{article12, title = {R4-SLAM: Toward Real-time, Robust, and Resource-Restricted Visual SLAM in Dynamic Environments}, author = {Zhijie Zhang and Cong Yang and Xue Qin and Wei Sui and Zeyd Boukhers and Hui Kong and Jie Liu and Zhijun Li}, journal = {IEEE Transactions on Instrumentation and Measurement}, year = {2025}, url = {}, publisher = {IEEE}, }
Visual Simultaneous Localization and Mapping (SLAM) in dynamic environments faces significant challenges due to moving objects that corrupt map accuracy and tracking stability. Existing methods either fail to efficiently handle dynamic scenes or require excessive computational resources, making them unsuitable for edge devices. In this paper, we propose R4-SLAM, a Real-time, Robust, and Resource-Restricted Visual SLAM system for dynamic environments. R4-SLAM introduces a semantic front-end decoupling strategy and semantic keyframe selection mechanism that effectively mitigates the impact of dynamic objects while maintaining high processing speed. Experimental results on the TUM RGB-D dataset demonstrate that R4-SLAM achieves state-of-the-art performance in terms of tracking accuracy, processing speed, and computational efficiency, making it suitable for deployment on resource-constrained mobile and edge devices.

2024

Skeleton ground truth extraction: methodology, annotation tool and benchmarks

skeleton ground truth deep learning heuristic extraction diagnosticity hypothesis SkeView tool evaluation baseline CNN shape datasets image datasets

Cong Yang , Bipin Indurkhya , John See , Bo Gao , Yan Ke , Zeyd Boukhers , Zhenyu Yang , Marcin Grzegorzek

International Journal of Computer Vision 132 1219-1241 2024

PDF DOI
Yang, C., Indurkhya, B., See, J., Gao, B., Ke, Y., Boukhers, Z., Yang, Z., & Grzegorzek, M. (2023). Skeleton ground truth extraction: methodology, annotation tool and benchmarks. International Journal of Computer Vision, 132(4), 1219–1241. https://doi.org/10.1007/s11263-023-01926-3

Yang, Cong, , and Bipin Indurkhya, , and John See, , and Bo Gao, , and Yan Ke, , and Zeyd Boukhers, , and Zhenyu Yang, , and Marcin Grzegorzek. "Skeleton ground truth extraction: methodology, annotation tool and benchmarks." International Journal of Computer Vision, vol. 132, no. 4, 2024, pp. 1219-1241.

Yang, Cong, , and Bipin Indurkhya, , and John See, , and Bo Gao, , and Yan Ke, , and Zeyd Boukhers, , and Zhenyu Yang, , and Marcin Grzegorzek. "Skeleton ground truth extraction: methodology, annotation tool and benchmarks." International Journal of Computer Vision 132, no. 4 (2024): 1219-1241.
@article{article1, title = {Skeleton ground truth extraction: methodology, annotation tool and benchmarks}, author = {Cong Yang and Bipin Indurkhya and John See and Bo Gao and Yan Ke and Zeyd Boukhers and Zhenyu Yang and Marcin Grzegorzek}, journal = {International Journal of Computer Vision}, volume = {132}, number = {4}, pages = {1219-1241}, year = {2024}, doi = {10.1007/s11263-023-01926-3}, url = {https://link.springer.com/content/pdf/10.1007/s11263-023-01926-3.pdf}, publisher = {Springer Science and Business Media LLC}, }
Skeleton Ground Truth (GT) is critical to the success of supervised skeleton extraction methods, especially with the popularity of deep learning techniques. Furthermore, we see skeleton GTs used not only for training skeleton detectors with Convolutional Neural Networks (CNN), but also for evaluating skeleton-related pruning and matching algorithms. However, most existing shape and image datasets suffer from the lack of skeleton GT and inconsistency of GT standards. As a result, it is difficult to evaluate and reproduce CNN-based skeleton detectors and algorithms on a fair basis. In this paper, we present a heuristic strategy for object skeleton GT extraction in binary shapes and natural images. Our strategy is built on an extended theory of diagnosticity hypothesis, which enables encoding human-in-the-loop GT extraction based on clues from the target’s context, simplicity, and completeness. Using this strategy, we developed a tool, SkeView, to generate skeleton GT of 17 existing shape and image datasets. The GTs are then structurally evaluated with representative methods to build viable baselines for fair comparisons. Experiments demonstrate that GTs generated by our strategy yield promising quality with respect to standard consistency, and also provide a balance between simplicity and completeness.
BladeView: Towards Automatic Wind Turbine Inspection with Unmanned Aerial Vehicle

Wind turbines Blades Inspection Drones Autonomous aerial vehicles Automation Usability Trajectory Parameter estimation Erosion Wind turbine blade inspection drone inspection unmanned aerial vehicle wind power

Cong Yang , Hua Zhou , Xun Liu , Yan Ke , Bo Gao , Marcin Grzegorzek , Zeyd Boukhers , Tao Chen , John See

IEEE Transactions on Automation Science and Engineering 22 7530-7545 2024

PDF DOI
Yang, C., Zhou, H., Liu, X., Ke, Y., Gao, B., Grzegorzek, M., Boukhers, Z., Chen, T., & See, J. (2024). BladeView: Towards Automatic Wind Turbine Inspection with Unmanned Aerial Vehicle. IEEE Transactions on Automation Science and Engineering, 22, 7530–7545. https://doi.org/10.1109/tase.2024.3464640

Yang, Cong, , and Hua Zhou, , and Xun Liu, , and Yan Ke, , and Bo Gao, , and Marcin Grzegorzek, , and Zeyd Boukhers, , and Tao Chen, , and John See. "BladeView: Towards Automatic Wind Turbine Inspection with Unmanned Aerial Vehicle." IEEE Transactions on Automation Science and Engineering, vol. 22, 2024, pp. 7530-7545.

Yang, Cong, , and Hua Zhou, , and Xun Liu, , and Yan Ke, , and Bo Gao, , and Marcin Grzegorzek, , and Zeyd Boukhers, , and Tao Chen, , and John See. "BladeView: Towards Automatic Wind Turbine Inspection with Unmanned Aerial Vehicle." IEEE Transactions on Automation Science and Engineering 22 (2024): 7530-7545.
@article{article9, title = {BladeView: Towards Automatic Wind Turbine Inspection with Unmanned Aerial Vehicle}, author = {Cong Yang and Hua Zhou and Xun Liu and Yan Ke and Bo Gao and Marcin Grzegorzek and Zeyd Boukhers and Tao Chen and John See}, journal = {IEEE Transactions on Automation Science and Engineering}, volume = {22}, pages = {7530-7545}, year = {2024}, doi = {10.1109/tase.2024.3464640}, url = {https://doi.org/10.1109/tase.2024.3464640}, publisher = {IEEE}, }
This paper presents a fully automatic method, BladeView, for drone-based wind turbine blade inspection using an Unmanned Aerial Vehicle (UAV). With the need for highly efficient blade inspection coupled with the rapid increase of wind turbines, existing methods provide limited automation on wind turbine parameter estimation, full blade coverage, and safety control. We introduce an Automatic Parameter Calculation (APC) algorithm and an Automatic Flight System (AFS) in BladeView to compute wind turbine parameters and inspection paths, respectively. Leveraging triangulation and linear fitting integration techniques, the APC automatically calculates the wind turbine parameters and estimates the relative angle and position between a drone and the turbine. Furthermore, with dynamic path finding and B-spline optimization, the AFS plans a path covering 3 blades within specified flight corridors, in compliance with the turbine parameters obtained from APC. Thus, the proposed BladeView can properly ensure an inspection’s automation, coverage, safety, and smoothness. The efficiency and usability of BladeView are validated through 100,000 flight simulations in the Gazebo simulation environment and 9,239 field runs at various wind farms, including offshore, near-shore, deserts, mountainous areas, farmlands, and suburbs.
A non-enhanced CT based deep learning diagnostic system for high-risk COVID-19 infection in lung cancer patients

COVID-19 non-enhanced CT deep learning lung cancer diagnostic system KL-6 3D CNN pneumonia classification

Tianming Du , Yihao Sun , Xinghao Wang , Tao Jiang , Ning Xu , Zeyd Boukhers , Marcin Grzegorzek , Hongzan Sun , Chen Li

Frontiers in Medicine 11 1444708 2024

PDF DOI
Du, T., Sun, Y., Wang, X., Jiang, T., Xu, N., Boukhers, Z., Grzegorzek, M., Sun, H., & Li, C. (2024). A non-enhanced CT based deep learning diagnostic system for high-risk COVID-19 infection in lung cancer patients. Frontiers in Medicine, 11, 1444708. https://doi.org/10.3389/fmed.2024.1444708

Du, Tianming, , and Yihao Sun, , and Xinghao Wang, , and Tao Jiang, , and Ning Xu, , and Zeyd Boukhers, , and Marcin Grzegorzek, , and Hongzan Sun, , and Chen Li. "A non-enhanced CT based deep learning diagnostic system for high-risk COVID-19 infection in lung cancer patients." Frontiers in Medicine, vol. 11, 2024, pp. 1444708.

Du, Tianming, , and Yihao Sun, , and Xinghao Wang, , and Tao Jiang, , and Ning Xu, , and Zeyd Boukhers, , and Marcin Grzegorzek, , and Hongzan Sun, , and Chen Li. "A non-enhanced CT based deep learning diagnostic system for high-risk COVID-19 infection in lung cancer patients." Frontiers in Medicine 11 (2024): 1444708.
@article{article10, title = {A non-enhanced CT based deep learning diagnostic system for high-risk COVID-19 infection in lung cancer patients}, author = {Tianming Du and Yihao Sun and Xinghao Wang and Tao Jiang and Ning Xu and Zeyd Boukhers and Marcin Grzegorzek and Hongzan Sun and Chen Li}, journal = {Frontiers in Medicine}, volume = {11}, pages = {1444708}, year = {2024}, doi = {10.3389/fmed.2024.1444708}, url = {https://doi.org/10.3389/fmed.2024.1444708}, publisher = {Frontiers}, }
Pneumonia and lung cancer have a mutually reinforcing relationship. Lung cancer patients are prone to contracting COVID-19, with poorer prognoses. Additionally, COVID-19 infection can impact anticancer treatments for lung cancer patients. Developing an early diagnostic system for COVID-19 pneumonia can help improve the prognosis of lung cancer patients with COVID-19 infection. This study proposes a neural network for COVID-19 diagnosis based on non-enhanced CT scans, consisting of two 3D convolutional neural networks (CNN) connected in series to form two diagnostic modules. The first diagnostic module classifies COVID-19 pneumonia patients from other pneumonia patients, while the second diagnostic module distinguishes severe COVID-19 patients from ordinary COVID-19 patients. We also analyzed the correlation between the deep learning features of the two diagnostic modules and various laboratory parameters, including KL-6. The first diagnostic module achieved an accuracy of 0.9669 on the training set and 0.8884 on the test set, while the second diagnostic module achieved an accuracy of 0.9722 on the training set and 0.9184 on the test set. Strong correlation was observed between the deep learning parameters of the second diagnostic module and KL-6. Our neural network can differentiate between COVID-19 pneumonia and other pneumonias on CT images, while also distinguishing between ordinary COVID-19 patients and those with white lung.

2023

Knowledge guided multi-filter residual convolutional neural network for ICD coding from clinical text

ICD coding knowledge graph deep neural networks clinical text MIMIC-III KG-MultiResCNN Wikidata medical informatics

Zeyd Boukhers , Prantik Goswami , Jan Jürjens

Neural Computing and Applications 35 17633-17644 2023

PDF DOI
Boukhers, Z., Goswami, P., & Jürjens, J. (2023). Knowledge guided multi-filter residual convolutional neural network for ICD coding from clinical text. Neural Computing and Applications, 35(24), 17633–17644. https://doi.org/10.1007/s00521-023-08581-2

Boukhers, Zeyd, , and Prantik Goswami, , and Jan Jürjens. "Knowledge guided multi-filter residual convolutional neural network for ICD coding from clinical text." Neural Computing and Applications, vol. 35, no. 24, 2023, pp. 17633-17644.

Boukhers, Zeyd, , and Prantik Goswami, , and Jan Jürjens. "Knowledge guided multi-filter residual convolutional neural network for ICD coding from clinical text." Neural Computing and Applications 35, no. 24 (2023): 17633-17644.
@article{article2, title = {Knowledge guided multi-filter residual convolutional neural network for ICD coding from clinical text}, author = {Zeyd Boukhers and Prantik Goswami and Jan Jürjens}, journal = {Neural Computing and Applications}, volume = {35}, number = {24}, pages = {17633-17644}, year = {2023}, doi = {10.1007/s00521-023-08581-2}, url = {https://doi.org/10.1007/s00521-023-08581-2}, publisher = {Springer Nature}, }
One challenge often encountered when using Deep Neural Network models for automatic ICD coding is their potential inability to effectively handle unseen clinical texts, especially when these models are only trained on a limited number of examples. This is because these models rely solely on the patterns and relationships present in the training data, and may not be able to effectively incorporate additional knowledge about the relationships between medical entities. To address this issue, we introduce KG-MultiResCNN - Knowledge Guided Multi-filter Residual Convolutional Neural Network model, which combines training examples with external knowledge from the Wikidata Knowledge Graph (KG) in order to better capture the relationships between medical entities. The KG is a structured database that contains a wealth of information about various entities, including medical concepts and their relationships with one another. By incorporating this external knowledge into our model, we are able to improve its ability to predict ICD codes for new clinical texts. In our experiments with the MIMIC-III dataset, we found that the KG-MultiResCNN model significantly outperformed the baseline approaches. This demonstrates the effectiveness of using external knowledge, in addition to training examples, to improve the performance of deep learning models for automatic ICD coding.
Deep author name disambiguation using DBLP data

author name disambiguation DBLP co-author network neural network model bibliographic records digital libraries name ambiguity

Zeyd Boukhers , Nagaraj Bahubali Asundi

International Journal on Digital Libraries 25 431-441 2023

PDF DOI
Boukhers, Z., & Asundi, N. B. (2023). Deep author name disambiguation using DBLP data. International Journal on Digital Libraries, 25(3), 431–441. https://doi.org/10.1007/s00799-023-00361-6

Boukhers, Zeyd, , and Nagaraj Bahubali Asundi. "Deep author name disambiguation using DBLP data." International Journal on Digital Libraries, vol. 25, no. 3, 2023, pp. 431-441.

Boukhers, Zeyd, , and Nagaraj Bahubali Asundi. "Deep author name disambiguation using DBLP data." International Journal on Digital Libraries 25, no. 3 (2023): 431-441.
@article{article3, title = {Deep author name disambiguation using DBLP data}, author = {Zeyd Boukhers and Nagaraj Bahubali Asundi}, journal = {International Journal on Digital Libraries}, volume = {25}, number = {3}, pages = {431-441}, year = {2023}, doi = {10.1007/s00799-023-00361-6}, url = {https://doi.org/10.1007/s00799-023-00361-6}, publisher = {Springer Science and Business Media LLC}, }
In the academic world, the number of scientists grows every year and so does the number of authors sharing the same names. Consequently, it is challenging to assign newly published papers to their respective authors. Therefore, author name ambiguity is considered a critical open problem in digital libraries. This paper proposes an author name disambiguation approach that links author names to their real-world entities by leveraging their co-authors and domain of research. To this end, we use data collected from the DBLP repository that contains more than 5 million bibliographic records authored by around 2.6 million co-authors. Our approach first groups authors who share the same last names and same first name initials. The author within each group is identified by capturing the relation with his/her co-authors and area of research, represented by the titles of the validated publications of the corresponding author. To this end, we train a neural network model that learns from the representations of the co-authors and titles. We validated the effectiveness of our approach by conducting extensive experiments on a large dataset.

2022

COIN: Counterfactual Image Generation for VQA Interpretation

counterfactual images visual question answering interpretability user study model behaviour explainable AI computer vision NLP

Zeyd Boukhers , Timo Hartmann , Jan Jürjens

Sensors 22 2245 2022

PDF DOI
Boukhers, Z., Hartmann, T., & Jürjens, J. (2022). COIN: Counterfactual Image Generation for VQA Interpretation. Sensors, 22(6), 2245. https://doi.org/10.3390/s22062245

Boukhers, Zeyd, , and Timo Hartmann, , and Jan Jürjens. "COIN: Counterfactual Image Generation for VQA Interpretation." Sensors, vol. 22, no. 6, 2022, pp. 2245.

Boukhers, Zeyd, , and Timo Hartmann, , and Jan Jürjens. "COIN: Counterfactual Image Generation for VQA Interpretation." Sensors 22, no. 6 (2022): 2245.
@article{article4, title = {COIN: Counterfactual Image Generation for VQA Interpretation}, author = {Zeyd Boukhers and Timo Hartmann and Jan Jürjens}, journal = {Sensors}, volume = {22}, number = {6}, pages = {2245}, year = {2022}, doi = {10.3390/s22062245}, url = {https://doi.org/10.3390/s22062245}, publisher = {MDPI}, }
Due to the significant advancement of Natural Language Processing and Computer Vision-based models, Visual Question Answering (VQA) systems are becoming more intelligent and advanced. However, they are still error-prone when dealing with relatively complex questions. Therefore, it is important to understand the behaviour of the VQA models before adopting their results. In this paper, we introduce an interpretability approach for VQA models by generating counterfactual images. Specifically, the generated image is supposed to have the minimal possible change to the original image and leads the VQA model to give a different answer. In addition, our approach ensures that the generated image is realistic. Since quantitative metrics cannot be employed to evaluate the interpretability of the model, we carried out a user study to assess different aspects of our approach. In addition to interpreting the result of VQA models on single images, the obtained results and the discussion provides an extensive explanation of VQA models’ behaviour.

2020

Analyzing the Influence of Hyper-parameters and Regularizers of Topic Modeling in Terms of Renyi Entropy

Topic modeling Renyi entropy Regularization Hyper-parameters Latent Dirichlet Allocation Hierarchical Dirichlet Process Additive Regularization of Topic Models pLSA BigARTM statistical physics

Sergei Koltsov , Vera Ignatenko , Zeyd Boukhers , Steffen Staab

Entropy 22 394 2020

PDF DOI
Koltsov, S., Ignatenko, V., Boukhers, Z., & Staab, S. (2020). Analyzing the Influence of Hyper-parameters and Regularizers of Topic Modeling in Terms of Renyi Entropy. Entropy, 22(4), 394. https://doi.org/10.3390/e22040394

Koltsov, Sergei, , and Vera Ignatenko, , and Zeyd Boukhers, , and Steffen Staab. "Analyzing the Influence of Hyper-parameters and Regularizers of Topic Modeling in Terms of Renyi Entropy." Entropy, vol. 22, no. 4, 2020, pp. 394.

Koltsov, Sergei, , and Vera Ignatenko, , and Zeyd Boukhers, , and Steffen Staab. "Analyzing the Influence of Hyper-parameters and Regularizers of Topic Modeling in Terms of Renyi Entropy." Entropy 22, no. 4 (2020): 394.
@article{article5, title = {Analyzing the Influence of Hyper-parameters and Regularizers of Topic Modeling in Terms of Renyi Entropy}, author = {Sergei Koltsov and Vera Ignatenko and Zeyd Boukhers and Steffen Staab}, journal = {Entropy}, volume = {22}, number = {4}, pages = {394}, year = {2020}, doi = {10.3390/e22040394}, url = {https://doi.org/10.3390/e22040394}, publisher = {MDPI}, }
Topic modeling is a popular technique for clustering large collections of text documents. A variety of different types of regularization is implemented in topic modeling. In this paper, we propose a novel approach for analyzing the influence of different regularization types on results of topic modeling. Based on Renyi entropy, this approach is inspired by the concepts from statistical physics, where an inferred topical structure of a collection can be considered an information statistical system residing in a non-equilibrium state. By testing our approach on four models—Probabilistic Latent Semantic Analysis (pLSA), Additive Regularization of Topic Models (BigARTM), Latent Dirichlet Allocation (LDA) with Gibbs sampling, LDA with variational inference (VLDA)—we, first of all, show that the minimum of Renyi entropy coincides with the "true" number of topics, as determined in two labelled collections. Simultaneously, we find that Hierarchical Dirichlet Process (HDP) model as a well-known approach for topic number optimization fails to detect such optimum. Next, we demonstrate that large values of the regularization coefficient in BigARTM significantly shift the minimum of entropy from the topic number optimum, which effect is not observed for hyper-parameters in LDA with Gibbs sampling. We conclude that regularization may introduce unpredictable distortions into topic models that need further research.

2019

Fractal approach for determining the optimal number of topics in the field of topic modeling

topic modeling fractal approach optimal topic number statistical physics text analysis Renyi entropy

Sergei Koltsov , Vera Ignatenko , Zeyd Boukhers , Steffen Staab

Journal of Physics: Conference Series 1163 012025 2019

PDF DOI
Koltsov, S., Ignatenko, V., Boukhers, Z., & Staab, S. (2019). Fractal approach for determining the optimal number of topics in the field of topic modeling. Journal of Physics: Conference Series, 1163, 012025. https://doi.org/10.1088/1742-6596/1163/1/012025

Koltsov, Sergei, , and Vera Ignatenko, , and Zeyd Boukhers, , and Steffen Staab. "Fractal approach for determining the optimal number of topics in the field of topic modeling." Journal of Physics: Conference Series, vol. 1163, 2019, pp. 012025.

Koltsov, Sergei, , and Vera Ignatenko, , and Zeyd Boukhers, , and Steffen Staab. "Fractal approach for determining the optimal number of topics in the field of topic modeling." Journal of Physics: Conference Series 1163 (2019): 012025.
@article{article6, title = {Fractal approach for determining the optimal number of topics in the field of topic modeling}, author = {Sergei Koltsov and Vera Ignatenko and Zeyd Boukhers and Steffen Staab}, journal = {Journal of Physics: Conference Series}, volume = {1163}, pages = {012025}, year = {2019}, doi = {10.1088/1742-6596/1163/1/012025}, url = {https://doi.org/10.1088/1742-6596/1163/1/012025}, publisher = {IOP Publishing}, }
The paper presents a fractal approach for determining the optimal number of topics in topic modeling. The method is based on the fractal properties of topical structures, using fractal measures to identify the optimal number of topics in a collection of text documents. The approach is validated on standard text corpora and demonstrates competitive performance compared to existing methods for optimal topic number estimation, offering a new perspective from the field of statistical physics applied to topic modeling.

2017

Example-based 3D trajectories of objects from 2D videos

3D trajectory extraction monocular camera 2D video human activity analysis example-based learning trajectory model object tracking

Zeyd Boukhers , Kimiaki Shirahama , Marcin Grzegorzek

IEEE Transactions on Circuits and Systems for Video Technology 28 2246-2260 2017

PDF DOI
Boukhers, Z., Shirahama, K., & Grzegorzek, M. (2017). Example-based 3D trajectories of objects from 2D videos. IEEE Transactions on Circuits and Systems for Video Technology, 28(9), 2246–2260. https://doi.org/10.1109/tcsvt.2017.2727963

Boukhers, Zeyd, , and Kimiaki Shirahama, , and Marcin Grzegorzek. "Example-based 3D trajectories of objects from 2D videos." IEEE Transactions on Circuits and Systems for Video Technology, vol. 28, no. 9, 2017, pp. 2246-2260.

Boukhers, Zeyd, , and Kimiaki Shirahama, , and Marcin Grzegorzek. "Example-based 3D trajectories of objects from 2D videos." IEEE Transactions on Circuits and Systems for Video Technology 28, no. 9 (2017): 2246-2260.
@article{article7, title = {Example-based 3D trajectories of objects from 2D videos}, author = {Zeyd Boukhers and Kimiaki Shirahama and Marcin Grzegorzek}, journal = {IEEE Transactions on Circuits and Systems for Video Technology}, volume = {28}, number = {9}, pages = {2246-2260}, year = {2017}, doi = {10.1109/tcsvt.2017.2727963}, url = {https://doi.org/10.1109/tcsvt.2017.2727963}, publisher = {IEEE}, }
In this paper, we propose an example-based approach to extract 3D trajectories of objects from 2D videos recorded by monocular moving cameras. Given a few example trajectories, our approach is able to learn a trajectory model that represents typical motions of objects in the scene. Using this model, the 3D positions of objects are estimated by solving an optimization problem that minimizes the difference between the observed 2D positions and the projected 3D positions. Experiments on challenging datasets demonstrate the effectiveness of our approach for human activity analysis in 3D space.
Less Restrictive Camera Odometry from Monocular Camera

camera odometry monocular camera depth estimation optical flow visual odometry SLAM self-localization

Zeyd Boukhers , Kimiaki Shirahama , Marcin Grzegorzek

Multimedia Tools and Applications 77 16199-16222 2017

PDF DOI
Boukhers, Z., Shirahama, K., & Grzegorzek, M. (2017). Less Restrictive Camera Odometry from Monocular Camera. Multimedia Tools and Applications, 77(13), 16199–16222. https://doi.org/10.1007/s11042-017-5195-7

Boukhers, Zeyd, , and Kimiaki Shirahama, , and Marcin Grzegorzek. "Less Restrictive Camera Odometry from Monocular Camera." Multimedia Tools and Applications, vol. 77, no. 13, 2017, pp. 16199-16222.

Boukhers, Zeyd, , and Kimiaki Shirahama, , and Marcin Grzegorzek. "Less Restrictive Camera Odometry from Monocular Camera." Multimedia Tools and Applications 77, no. 13 (2017): 16199-16222.
@article{article8, title = {Less Restrictive Camera Odometry from Monocular Camera}, author = {Zeyd Boukhers and Kimiaki Shirahama and Marcin Grzegorzek}, journal = {Multimedia Tools and Applications}, volume = {77}, number = {13}, pages = {16199-16222}, year = {2017}, doi = {10.1007/s11042-017-5195-7}, url = {https://doi.org/10.1007/s11042-017-5195-7}, publisher = {Springer International Publishing}, }
Camera odometry estimates the motion of a camera from image sequences. Most existing approaches rely on restrictive assumptions about the camera motion or the scene structure, which limits their applicability in real-world scenarios. In this paper, we propose a less restrictive camera odometry approach from monocular camera that relaxes these assumptions. Our approach integrates depth estimation and optical flow to robustly estimate camera motion without requiring a known camera calibration or prior knowledge about the scene. Experiments on standard benchmark datasets demonstrate that our approach achieves competitive performance compared to state-of-the-art methods.

Conference Papers

2026

Unleashing Semantic and Geometric Priors for 3D Scene Completion

3D scene completion semantic priors geometric priors deep learning 3D reconstruction point cloud AAAI

Shiyuan Chen , Wei Sui , Bohao Zhang , Zeyd Boukhers , John See , Cong Yang

Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence 2026

PDF
Chen, S., Sui, W., Zhang, B., Boukhers, Z., See, J., & Yang, C. (2026). Unleashing Semantic and Geometric Priors for 3D Scene Completion. Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence.

Chen, Shiyuan, , and Wei Sui, , and Bohao Zhang, , and Zeyd Boukhers, , and John See, , and Cong Yang. "Unleashing Semantic and Geometric Priors for 3D Scene Completion." Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence, 2026.

Chen, Shiyuan, , and Wei Sui, , and Bohao Zhang, , and Zeyd Boukhers, , and John See, , and Cong Yang. "Unleashing Semantic and Geometric Priors for 3D Scene Completion." Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence (2026).
@inproceedings{Chen2026unleashing, title = {Unleashing Semantic and Geometric Priors for 3D Scene Completion}, author = {Shiyuan Chen and Wei Sui and Bohao Zhang and Zeyd Boukhers and John See and Cong Yang}, booktitle = {Proceedings of the 40th Annual AAAI Conference on Artificial Intelligence}, year = {2026}, url = {}, }
3D scene completion aims to reconstruct complete 3D scenes from partial observations. Existing methods often struggle with large missing regions and lack of structural understanding. This paper presents a method that unleashes semantic and geometric priors for 3D scene completion. By incorporating pre-trained semantic knowledge and geometric constraints into the completion process, our approach generates more realistic and structurally coherent completions. Experiments on standard benchmarks demonstrate state-of-the-art performance across diverse scene types.
Enhancing RL Generalizability in Robotics through SHAP Analysis of Algorithms and Hyperparameters

reinforcement learning SHAP robotics generalization hyperparameters explainability sim-to-real transfer pattern recognition

Lingxiao Kong , Cong Yang , Oya Beyan , Zeyd Boukhers

28th International Conference on Pattern Recognition (ICPR 2026) 2026

PDF
Kong, L., Yang, C., Beyan, O., & Boukhers, Z. (2026). Enhancing RL Generalizability in Robotics through SHAP Analysis of Algorithms and Hyperparameters. 28th International Conference on Pattern Recognition (ICPR 2026).

Kong, Lingxiao, , and Cong Yang, , and Oya Beyan, , and Zeyd Boukhers. "Enhancing RL Generalizability in Robotics through SHAP Analysis of Algorithms and Hyperparameters." 28th International Conference on Pattern Recognition (ICPR 2026), 2026.

Kong, Lingxiao, , and Cong Yang, , and Oya Beyan, , and Zeyd Boukhers. "Enhancing RL Generalizability in Robotics through SHAP Analysis of Algorithms and Hyperparameters." 28th International Conference on Pattern Recognition (ICPR 2026) (2026).
@inproceedings{Kong2026enhancing, title = {Enhancing RL Generalizability in Robotics through SHAP Analysis of Algorithms and Hyperparameters}, author = {Lingxiao Kong and Cong Yang and Oya Beyan and Zeyd Boukhers}, booktitle = {28th International Conference on Pattern Recognition (ICPR 2026)}, year = {2026}, url = {}, }
Reinforcement learning (RL) algorithms for robotics often fail to generalize across different environments due to sensitivity to algorithm choices and hyperparameter settings. This paper analyzes RL generalizability in robotics through SHAP (SHapley Additive exPlanations) values to identify the contribution of different algorithms and hyperparameters to sim-to-real transfer performance. By quantifying the impact of these factors, the approach provides actionable insights to improve RL policy robustness and generalization in real-world robotic deployment.

2025

From Theory to Practice: Demonstrators of FAIR Data Spaces Across Different Sectors

FAIR Data Spaces data economy data sovereignty healthcare manufacturing research infrastructure Gaia-X NFDI

Nikolaus Glombiewski , Zeyd Boukhers , Christian Beilschmidt , Johannes Drönner , Michael Mattig , Artur Piet , Robert Pietrzynski , Mehrshad Jaberansary , Macedo Maia , Sebastian Beyvers , Yeliz Üçer Yediel , Muhammad Hamza Akhtar , Heiner Oberkampf , Jonathan Hartman , Christoph Lange

The 40th ACM/SIGAPP Symposium On Applied Computing (SAC 2025) 1-4 2025

PDF DOI
Glombiewski, N., Boukhers, Z., Beilschmidt, C., Drönner, J., Mattig, M., Piet, A., Pietrzynski, R., Jaberansary, M., Maia, M., Beyvers, S., Yediel, Y. Ü., Akhtar, M. H., Oberkampf, H., Hartman, J., & Lange, C. (2025). From Theory to Practice: Demonstrators of FAIR Data Spaces Across Different Sectors. The 40th ACM/SIGAPP Symposium On Applied Computing (SAC 2025), 1–4. https://doi.org/10.1145/3672608.3707987

Glombiewski, Nikolaus, , and Zeyd Boukhers, , and Christian Beilschmidt, , and Johannes Drönner, , and Michael Mattig, , and Artur Piet, , and Robert Pietrzynski, , and Mehrshad Jaberansary, , and Macedo Maia, , and Sebastian Beyvers, , and Yeliz Üçer Yediel, , and Muhammad Hamza Akhtar, , and Heiner Oberkampf, , and Jonathan Hartman, , and Christoph Lange. "From Theory to Practice: Demonstrators of FAIR Data Spaces Across Different Sectors." The 40th ACM/SIGAPP Symposium On Applied Computing (SAC 2025), 2025, pp. 1-4.

Glombiewski, Nikolaus, , and Zeyd Boukhers, , and Christian Beilschmidt, , and Johannes Drönner, , and Michael Mattig, , and Artur Piet, , and Robert Pietrzynski, , and Mehrshad Jaberansary, , and Macedo Maia, , and Sebastian Beyvers, , and Yeliz Üçer Yediel, , and Muhammad Hamza Akhtar, , and Heiner Oberkampf, , and Jonathan Hartman, , and Christoph Lange. "From Theory to Practice: Demonstrators of FAIR Data Spaces Across Different Sectors." The 40th ACM/SIGAPP Symposium On Applied Computing (SAC 2025) (2025): 1-4.
@inproceedings{Glombiewski2025from, title = {From Theory to Practice: Demonstrators of FAIR Data Spaces Across Different Sectors}, author = {Nikolaus Glombiewski and Zeyd Boukhers and Christian Beilschmidt and Johannes Drönner and Michael Mattig and Artur Piet and Robert Pietrzynski and Mehrshad Jaberansary and Macedo Maia and Sebastian Beyvers and Yeliz Üçer Yediel and Muhammad Hamza Akhtar and Heiner Oberkampf and Jonathan Hartman and Christoph Lange}, booktitle = {The 40th ACM/SIGAPP Symposium On Applied Computing (SAC 2025)}, pages = {1-4}, year = {2025}, doi = {10.1145/3672608.3707987}, url = {https://doi.org/10.1145/3672608.3707987}, publisher = {ACM}, }
FAIR Data Spaces represent a key concept in the European data economy, enabling trusted and sovereign data sharing across sectors. This paper presents demonstrators of FAIR Data Spaces implemented across different sectors including healthcare, manufacturing, and research. The demonstrators showcase how the abstract concepts of FAIR Data Spaces can be instantiated in real-world scenarios, providing practical guidance for organizations seeking to participate in the data economy while maintaining data sovereignty.
ELMTEX: Fine-tuning large language models for structured clinical information extraction. A case study on clinical reports

ELMTEX fine-tuning large language models structured clinical information extraction clinical reports radiology NLP medical AI

Aynur Guluzade , Naguib Heiba , Zeyd Boukhers , Florim Hamiti , Jahid Hasan Polash , Yehya Mohamad , Carlos A Velasco

International Conference on Artificial Intelligence in Medicine (AIME 2025) 181-185 2025

PDF DOI
Guluzade, A., Heiba, N., Boukhers, Z., Hamiti, F., Polash, J. H., Mohamad, Y., & Velasco, C. A. (2025). ELMTEX: Fine-tuning large language models for structured clinical information extraction. A case study on clinical reports. International Conference on Artificial Intelligence in Medicine (AIME 2025), 181–185. https://doi.org/10.1007/978-3-031-95841-0_34

Guluzade, Aynur, , and Naguib Heiba, , and Zeyd Boukhers, , and Florim Hamiti, , and Jahid Hasan Polash, , and Yehya Mohamad, , and Carlos A Velasco. "ELMTEX: Fine-tuning large language models for structured clinical information extraction. A case study on clinical reports." International Conference on Artificial Intelligence in Medicine (AIME 2025), 2025, pp. 181-185.

Guluzade, Aynur, , and Naguib Heiba, , and Zeyd Boukhers, , and Florim Hamiti, , and Jahid Hasan Polash, , and Yehya Mohamad, , and Carlos A Velasco. "ELMTEX: Fine-tuning large language models for structured clinical information extraction. A case study on clinical reports." International Conference on Artificial Intelligence in Medicine (AIME 2025) (2025): 181-185.
@inproceedings{Guluzade2025elmtex, title = {ELMTEX: Fine-tuning large language models for structured clinical information extraction. A case study on clinical reports}, author = {Aynur Guluzade and Naguib Heiba and Zeyd Boukhers and Florim Hamiti and Jahid Hasan Polash and Yehya Mohamad and Carlos A Velasco}, booktitle = {International Conference on Artificial Intelligence in Medicine (AIME 2025)}, pages = {181-185}, year = {2025}, doi = {10.1007/978-3-031-95841-0_34}, url = {https://doi.org/10.1007/978-3-031-95841-0_34}, publisher = {Springer Nature Switzerland}, }
Clinical reports contain rich structured information that is valuable for medical research and decision support. However, automatically extracting structured information from free-text clinical reports remains challenging due to the diversity of clinical terminology and report formats. This paper presents ELMTEX, a fine-tuning approach for large language models (LLMs) tailored to structured clinical information extraction. ELMTEX employs a task-specific fine-tuning strategy combined with domain adaptation to extract information such as diagnoses, medications, procedures, and temporal relations from clinical reports. A case study on radiology reports demonstrates that ELMTEX achieves state-of-the-art performance on structured information extraction benchmarks.
Future-Proofing Heritage with ARGUS: A Multimodal Digital Twin Approach for Sustainable Preservation

digital heritage digital twin ARGUS 3D scanning photogrammetry cultural heritage archaeological sites sustainable preservation multimodal

George Pavlidis , Anestis Koutsoudis , Despina Tsiafaki , Vasileios Sevetlidis , Vasileios Arampatsakis , Apostolos Sarris , Miltiades Polidorou , Victor Klinkenberg , Zeyd Boukhers , Lingxiao Kong , Emeri Farinetti , Fernando Moreno-Navarro , Ioannis Kakogiannos , Sofia Aparicio , Javier Ortega , Fernando Ramonet , Athos Agapiou , Stylianos Hadjipetrou , Kyriakos Michaelidis , Stavros Patsalidis , Phaedon Kyriaкidis , Apostolos Papakonstandinou , Demetrios Athanasoulis , Christos Maris

International Conference on Computer Applications & Quantitative Methods in Archaeology 2025: "Digital Horizons" 2025

PDF DOI
Pavlidis, G., Koutsoudis, A., Tsiafaki, D., Sevetlidis, V., Arampatsakis, V., Sarris, A., Polidorou, M., Klinkenberg, V., Boukhers, Z., Kong, L., Farinetti, E., Moreno-Navarro, F., Kakogiannos, I., Aparicio, S., Ortega, J., Ramonet, F., Agapiou, A., Hadjipetrou, S., Michaelidis, K., … Maris, C. (2025). Future-Proofing Heritage with ARGUS: A Multimodal Digital Twin Approach for Sustainable Preservation. International Conference on Computer Applications & Quantitative Methods in Archaeology 2025: "Digital Horizons". https://doi.org/10.24072/pci.archaeo.100624.d1

Pavlidis, George, , and Anestis Koutsoudis, , and Despina Tsiafaki, , and Vasileios Sevetlidis, , and Vasileios Arampatsakis, , and Apostolos Sarris, , and Miltiades Polidorou, , and Victor Klinkenberg, , and Zeyd Boukhers, , and Lingxiao Kong, , and Emeri Farinetti, , and Fernando Moreno-Navarro, , and Ioannis Kakogiannos, , and Sofia Aparicio, , and Javier Ortega, , and Fernando Ramonet, , and Athos Agapiou, , and Stylianos Hadjipetrou, , and Kyriakos Michaelidis, , and Stavros Patsalidis, , and Phaedon Kyriaкidis, , and Apostolos Papakonstandinou, , and Demetrios Athanasoulis, , and Christos Maris. "Future-Proofing Heritage with ARGUS: A Multimodal Digital Twin Approach for Sustainable Preservation." International Conference on Computer Applications & Quantitative Methods in Archaeology 2025: "Digital Horizons", 2025.

Pavlidis, George, , and Anestis Koutsoudis, , and Despina Tsiafaki, , and Vasileios Sevetlidis, , and Vasileios Arampatsakis, , and Apostolos Sarris, , and Miltiades Polidorou, , and Victor Klinkenberg, , and Zeyd Boukhers, , and Lingxiao Kong, , and Emeri Farinetti, , and Fernando Moreno-Navarro, , and Ioannis Kakogiannos, , and Sofia Aparicio, , and Javier Ortega, , and Fernando Ramonet, , and Athos Agapiou, , and Stylianos Hadjipetrou, , and Kyriakos Michaelidis, , and Stavros Patsalidis, , and Phaedon Kyriaкidis, , and Apostolos Papakonstandinou, , and Demetrios Athanasoulis, , and Christos Maris. "Future-Proofing Heritage with ARGUS: A Multimodal Digital Twin Approach for Sustainable Preservation." International Conference on Computer Applications & Quantitative Methods in Archaeology 2025: "Digital Horizons" (2025).
@inproceedings{Pavlidis2025future, title = {Future-Proofing Heritage with ARGUS: A Multimodal Digital Twin Approach for Sustainable Preservation}, author = {George Pavlidis and Anestis Koutsoudis and Despina Tsiafaki and Vasileios Sevetlidis and Vasileios Arampatsakis and Apostolos Sarris and Miltiades Polidorou and Victor Klinkenberg and Zeyd Boukhers and Lingxiao Kong and Emeri Farinetti and Fernando Moreno-Navarro and Ioannis Kakogiannos and Sofia Aparicio and Javier Ortega and Fernando Ramonet and Athos Agapiou and Stylianos Hadjipetrou and Kyriakos Michaelidis and Stavros Patsalidis and Phaedon Kyriaкidis and Apostolos Papakonstandinou and Demetrios Athanasoulis and Christos Maris}, booktitle = {International Conference on Computer Applications & Quantitative Methods in Archaeology 2025: "Digital Horizons"}, year = {2025}, doi = {10.24072/pci.archaeo.100624.d1}, url = {https://doi.org/10.24072/pci.archaeo.100624.d1}, publisher = {Peer Community In}, }
Digital heritage preservation requires robust approaches to manage the long-term accessibility and sustainability of cultural heritage data. This paper presents ARGUS, a multimodal digital twin approach for sustainable preservation of archaeological and cultural heritage sites. ARGUS integrates 3D scanning, photogrammetry, multispectral imaging, and semantic web technologies to create comprehensive digital twins of heritage assets. The paper discusses the architecture, implementation, and evaluation of ARGUS demonstrators at multiple heritage sites across Europe.
Q-Chain: A Causal-Aware Framework for Structural and Educational Question Generation

question generation causal reasoning educational AI NLP knowledge graphs reading comprehension Q-Chain

Junqi Xu , Lvcheng Wang , Zeyd Boukhers , Bipin Indurkhya , Cong Yang

The 15th ACM International Conference on Multimedia Retrieval 2083-2087 2025

PDF DOI
Xu, J., Wang, L., Boukhers, Z., Indurkhya, B., & Yang, C. (2025). Q-Chain: A Causal-Aware Framework for Structural and Educational Question Generation. The 15th ACM International Conference on Multimedia Retrieval, 2083–2087. https://doi.org/10.1145/3731715.3733492

Xu, Junqi, , and Lvcheng Wang, , and Zeyd Boukhers, , and Bipin Indurkhya, , and Cong Yang. "Q-Chain: A Causal-Aware Framework for Structural and Educational Question Generation." The 15th ACM International Conference on Multimedia Retrieval, 2025, pp. 2083-2087.

Xu, Junqi, , and Lvcheng Wang, , and Zeyd Boukhers, , and Bipin Indurkhya, , and Cong Yang. "Q-Chain: A Causal-Aware Framework for Structural and Educational Question Generation." The 15th ACM International Conference on Multimedia Retrieval (2025): 2083-2087.
@inproceedings{Xu2025q-chain, title = {Q-Chain: A Causal-Aware Framework for Structural and Educational Question Generation}, author = {Junqi Xu and Lvcheng Wang and Zeyd Boukhers and Bipin Indurkhya and Cong Yang}, booktitle = {The 15th ACM International Conference on Multimedia Retrieval}, pages = {2083-2087}, year = {2025}, doi = {10.1145/3731715.3733492}, url = {https://doi.org/10.1145/3731715.3733492}, publisher = {ACM}, }
Automatic question generation (QG) is an important task for educational applications and reading comprehension systems. Existing approaches often generate questions that lack causal reasoning or fail to capture the structural relationships within text. This paper presents Q-Chain, a causal-aware framework for structural and educational question generation. Q-Chain leverages causal graphs extracted from text to guide question generation, producing questions that probe causal relationships and promote deeper understanding. Experiments on educational datasets demonstrate that Q-Chain outperforms baselines in question quality, diversity, and educational value.
A unified framework for cultural heritage data historicity and migration: The ARGUS approach

cultural heritage data historicity data migration ARGUS semantic web linked data provenance digital humanities

Lingxiao Kong , Apostolos Sarris , Miltiades Polidorou , Victor Klingenberg , Vasilis Sevetlidis , Vasilis Arampatsakis , George Pavlidis , Cong Yang , Zeyd Boukhers

IEEE International Conference on Cyber Humanities 2025
Kong, L., Sarris, A., Polidorou, M., Klingenberg, V., Sevetlidis, V., Arampatsakis, V., Pavlidis, G., Yang, C., & Boukhers, Z. (2025). A unified framework for cultural heritage data historicity and migration: The ARGUS approach. IEEE International Conference on Cyber Humanities.

Kong, Lingxiao, , and Apostolos Sarris, , and Miltiades Polidorou, , and Victor Klingenberg, , and Vasilis Sevetlidis, , and Vasilis Arampatsakis, , and George Pavlidis, , and Cong Yang, , and Zeyd Boukhers. "A unified framework for cultural heritage data historicity and migration: The ARGUS approach." IEEE International Conference on Cyber Humanities, 2025.

Kong, Lingxiao, , and Apostolos Sarris, , and Miltiades Polidorou, , and Victor Klingenberg, , and Vasilis Sevetlidis, , and Vasilis Arampatsakis, , and George Pavlidis, , and Cong Yang, , and Zeyd Boukhers. "A unified framework for cultural heritage data historicity and migration: The ARGUS approach." IEEE International Conference on Cyber Humanities (2025).
@inproceedings{Kong2025a, title = {A unified framework for cultural heritage data historicity and migration: The ARGUS approach}, author = {Lingxiao Kong and Apostolos Sarris and Miltiades Polidorou and Victor Klingenberg and Vasilis Sevetlidis and Vasilis Arampatsakis and George Pavlidis and Cong Yang and Zeyd Boukhers}, booktitle = {IEEE International Conference on Cyber Humanities}, year = {2025}, }
Cultural heritage data preservation faces challenges around data historicity - tracking how information about heritage assets changes over time - and data migration across heterogeneous digital systems. This paper presents the ARGUS approach, a unified framework for cultural heritage data historicity and migration. ARGUS employs semantic web technologies and linked data principles to maintain provenance and version information about heritage data, enabling reliable migration between systems while preserving the historical context of information updates.
EMORL: Ensemble multi-objective reinforcement learning for efficient and flexible LLM fine-tuning

reinforcement learning large language models multi-objective optimization fine-tuning ensemble learning LLM alignment counselor reflection

Lingxiao Kong , Cong Yang , Susanne Neufang , Oya Deniz Beyan , Zeyd Boukhers

Proceedings of the 26th Annual Meeting of the Special Interest Group on Discourse and Dialogue 417–430 2025

PDF
Kong, L., Yang, C., Neufang, S., Beyan, O. D., & Boukhers, Z. (2025). EMORL: Ensemble multi-objective reinforcement learning for efficient and flexible LLM fine-tuning. Proceedings of the 26th Annual Meeting of the Special Interest Group on Discourse and Dialogue, 417–430. https://aclanthology.org/2025.sigdial-1.33/

Kong, Lingxiao, , and Cong Yang, , and Susanne Neufang, , and Oya Deniz Beyan, , and Zeyd Boukhers. "EMORL: Ensemble multi-objective reinforcement learning for efficient and flexible LLM fine-tuning." Proceedings of the 26th Annual Meeting of the Special Interest Group on Discourse and Dialogue, 2025, pp. 417–430.

Kong, Lingxiao, , and Cong Yang, , and Susanne Neufang, , and Oya Deniz Beyan, , and Zeyd Boukhers. "EMORL: Ensemble multi-objective reinforcement learning for efficient and flexible LLM fine-tuning." Proceedings of the 26th Annual Meeting of the Special Interest Group on Discourse and Dialogue (2025): 417–430.
@inproceedings{Kong2025emorl, title = {EMORL: Ensemble multi-objective reinforcement learning for efficient and flexible LLM fine-tuning}, author = {Lingxiao Kong and Cong Yang and Susanne Neufang and Oya Deniz Beyan and Zeyd Boukhers}, booktitle = {Proceedings of the 26th Annual Meeting of the Special Interest Group on Discourse and Dialogue}, pages = {417–430}, year = {2025}, url = {https://aclanthology.org/2025.sigdial-1.33/}, publisher = {Association for Computational Linguistics}, }
Recent advances in reinforcement learning (RL) for large language model (LLM) fine-tuning show promise in addressing multi-objective tasks but still face significant challenges, including competing objective balancing, low training efficiency, poor scalability, and limited explainability. Leveraging ensemble learning principles, we introduce an Ensemble Multi-Objective RL (EMORL) framework that fine-tunes multiple models with individual objectives while optimizing their aggregation after the fine-tuning to improve efficiency and flexibility. Our method is the first to aggregate the hidden states of individual models, incorporating contextual information from multiple objectives. This approach is supported by a hierarchical grid search algorithm that identifies optimal weighted combinations. We evaluate EMORL on counselor reflection generation tasks, using text classification models to score the generations and provide rewards during RL fine-tuning. Through comprehensive experiments on the PAIR and Psych8k datasets, we demonstrate the advantages of EMORL against existing baselines: significantly lower and more stable training consumption, improved scalability and explainability, and comparable performance across multiple objectives.
Exploring the limits of model compression in LLMs: a knowledge distillation study on QA tasks

knowledge distillation large language models model compression question answering NLP efficient LLMs

Joyeeta Datta , Niclas Doll , Qusai Ramadan , Zeyd Boukhers

Proceedings of the 26th Annual Meeting of the Special Interest Group on Discourse and Dialogue 477–483 2025

PDF
Datta, J., Doll, N., Ramadan, Q., & Boukhers, Z. (2025). Exploring the limits of model compression in LLMs: a knowledge distillation study on QA tasks. Proceedings of the 26th Annual Meeting of the Special Interest Group on Discourse and Dialogue, 477–483. https://aclanthology.org/2025.sigdial-1.39/

Datta, Joyeeta, , and Niclas Doll, , and Qusai Ramadan, , and Zeyd Boukhers. "Exploring the limits of model compression in LLMs: a knowledge distillation study on QA tasks." Proceedings of the 26th Annual Meeting of the Special Interest Group on Discourse and Dialogue, 2025, pp. 477–483.

Datta, Joyeeta, , and Niclas Doll, , and Qusai Ramadan, , and Zeyd Boukhers. "Exploring the limits of model compression in LLMs: a knowledge distillation study on QA tasks." Proceedings of the 26th Annual Meeting of the Special Interest Group on Discourse and Dialogue (2025): 477–483.
@inproceedings{Datta2025exploring, title = {Exploring the limits of model compression in LLMs: a knowledge distillation study on QA tasks}, author = {Joyeeta Datta and Niclas Doll and Qusai Ramadan and Zeyd Boukhers}, booktitle = {Proceedings of the 26th Annual Meeting of the Special Interest Group on Discourse and Dialogue}, pages = {477–483}, year = {2025}, url = {https://aclanthology.org/2025.sigdial-1.39/}, publisher = {Association for Computational Linguistics}, }
Large Language Models (LLMs) have shown outstanding performance across a range of NLP tasks, but their computational demands hinder deployment in real-world, resource-constrained environments. This work investigates the extent to which LLMs can be compressed using knowledge distillation (KD) while maintaining strong performance on question answering (QA) tasks. We evaluate student models distilled from the Pythia and Qwen2.5 families on two QA benchmarks, SQuAD and MLQA, under zero-shot and one-shot prompting conditions. Results show that student models retain over 90% of their teacher models’ performance while reducing parameter counts by up to 57.1%. Furthermore, one-shot prompting yields additional performance gains over zero-shot setups for both model families. These findings underscore the trade-off between model efficiency and task performance, demonstrating that KD, combined with minimal prompting, can yield compact yet capable QA systems suitable for real-world applications.
FAIR Data Assessment Using LLMs: The Fair-Way

FAIR data FAIRness assessment large language models research data management metadata evaluation open science

Anmol Sharma , Sulayman K. Sowe , Soo-Yon Kim , Sayed Hosseini , Fidan Limani , Zeyd Boukhers , Christoph Lange , Stefan Decker

The Conference on Information and Knowledge Management (CIKM) 5228-5232 2025

PDF DOI
Sharma, A., Sowe, S. K., Kim, S.-Y., Hosseini, S., Limani, F., Boukhers, Z., Lange, C., & Decker, S. (2025). FAIR Data Assessment Using LLMs: The Fair-Way. The Conference on Information and Knowledge Management (CIKM), 5228–5232. https://doi.org/10.1145/3746252.3760811

Sharma, Anmol, , and Sulayman K. Sowe, , and Soo-Yon Kim, , and Sayed Hosseini, , and Fidan Limani, , and Zeyd Boukhers, , and Christoph Lange, , and Stefan Decker. "FAIR Data Assessment Using LLMs: The Fair-Way." The Conference on Information and Knowledge Management (CIKM), 2025, pp. 5228-5232.

Sharma, Anmol, , and Sulayman K. Sowe, , and Soo-Yon Kim, , and Sayed Hosseini, , and Fidan Limani, , and Zeyd Boukhers, , and Christoph Lange, , and Stefan Decker. "FAIR Data Assessment Using LLMs: The Fair-Way." The Conference on Information and Knowledge Management (CIKM) (2025): 5228-5232.
@inproceedings{Sharma2025fair, title = {FAIR Data Assessment Using LLMs: The Fair-Way}, author = {Anmol Sharma and Sulayman K. Sowe and Soo-Yon Kim and Sayed Hosseini and Fidan Limani and Zeyd Boukhers and Christoph Lange and Stefan Decker}, booktitle = {The Conference on Information and Knowledge Management (CIKM)}, pages = {5228-5232}, year = {2025}, doi = {10.1145/3746252.3760811}, url = {https://doi.org/10.1145/3746252.3760811}, publisher = {ACM}, }
FAIR data assessment is essential for research data management and compliance with open science policies. Manual assessment of data FAIRness is time-consuming and requires domain expertise. This paper presents Fair-Way, an approach for automated FAIR data assessment using large language models (LLMs). Fair-Way prompts LLMs to evaluate metadata records against FAIR criteria, generating structured assessments with explanations. Experiments on diverse metadata repositories demonstrate that Fair-Way achieves strong agreement with expert assessments while being significantly faster.

2024

YawnNet: A Visual-Centric Approach for Yawning Detection

yawning detection driver drowsiness fatigue monitoring deep learning convolutional neural network facial analysis safety systems

Ruoxi Sun , Xinyu Yang , Cong Qian , Chenyu Zhu , Wei Sui , Zeyd Boukhers , Cong Yang

Proceedings of the 2024 International Conference on Multimedia Retrieval 1140-1144 2024

PDF DOI
Sun, R., Yang, X., Qian, C., Zhu, C., Sui, W., Boukhers, Z., & Yang, C. (2024). YawnNet: A Visual-Centric Approach for Yawning Detection. Proceedings of the 2024 International Conference on Multimedia Retrieval, 1140–1144. https://doi.org/10.1145/3652583.3657618

Sun, Ruoxi, , and Xinyu Yang, , and Cong Qian, , and Chenyu Zhu, , and Wei Sui, , and Zeyd Boukhers, , and Cong Yang. "YawnNet: A Visual-Centric Approach for Yawning Detection." Proceedings of the 2024 International Conference on Multimedia Retrieval, 2024, pp. 1140-1144.

Sun, Ruoxi, , and Xinyu Yang, , and Cong Qian, , and Chenyu Zhu, , and Wei Sui, , and Zeyd Boukhers, , and Cong Yang. "YawnNet: A Visual-Centric Approach for Yawning Detection." Proceedings of the 2024 International Conference on Multimedia Retrieval (2024): 1140-1144.
@inproceedings{Sun2024yawnet, title = {YawnNet: A Visual-Centric Approach for Yawning Detection}, author = {Ruoxi Sun and Xinyu Yang and Cong Qian and Chenyu Zhu and Wei Sui and Zeyd Boukhers and Cong Yang}, booktitle = {Proceedings of the 2024 International Conference on Multimedia Retrieval}, pages = {1140-1144}, year = {2024}, doi = {10.1145/3652583.3657618}, url = {https://doi.org/10.1145/3652583.3657618}, publisher = {ACM}, }
Yawning detection is an important task in driver drowsiness monitoring and fatigue assessment systems. Existing approaches often rely on facial landmarks or optical flow, which can be sensitive to occlusions and pose variations. This paper presents YawnNet, a visual-centric approach for yawning detection that focuses on appearance features extracted from the mouth region. YawnNet employs a deep convolutional neural network trained on a diverse dataset of yawning and non-yawning instances, achieving robust performance across different subjects, lighting conditions, and camera angles.
Moiré pattern detection: Stability and efficiency with evaluated loss function

Moiré pattern image quality artifact detection frequency domain pattern recognition image processing loss function

Zhuocheng Li , Xizhu Shen , Simin Luan , Shuwei Guo , Zeyd Boukhers , Yuyi Wang , Cong Yang

International Conference on Pattern Recognition 1-16 2024

PDF DOI
Li, Z., Shen, X., Luan, S., Guo, S., Boukhers, Z., Wang, Y., & Yang, C. (2024). Moiré pattern detection: Stability and efficiency with evaluated loss function. International Conference on Pattern Recognition, 1–16. https://doi.org/10.1007/978-3-031-78444-6_3

Li, Zhuocheng, , and Xizhu Shen, , and Simin Luan, , and Shuwei Guo, , and Zeyd Boukhers, , and Yuyi Wang, , and Cong Yang. "Moiré pattern detection: Stability and efficiency with evaluated loss function." International Conference on Pattern Recognition, 2024, pp. 1-16.

Li, Zhuocheng, , and Xizhu Shen, , and Simin Luan, , and Shuwei Guo, , and Zeyd Boukhers, , and Yuyi Wang, , and Cong Yang. "Moiré pattern detection: Stability and efficiency with evaluated loss function." International Conference on Pattern Recognition (2024): 1-16.
@inproceedings{Li2024moire, title = {Moiré pattern detection: Stability and efficiency with evaluated loss function}, author = {Zhuocheng Li and Xizhu Shen and Simin Luan and Shuwei Guo and Zeyd Boukhers and Yuyi Wang and Cong Yang}, booktitle = {International Conference on Pattern Recognition}, pages = {1-16}, year = {2024}, doi = {10.1007/978-3-031-78444-6_3}, url = {https://doi.org/10.1007/978-3-031-78444-6_3}, publisher = {Springer Nature Switzerland}, }
Moiré patterns are interference artifacts that appear when digital images capture fine periodic structures such as screens or textiles. Detecting and removing moiré patterns is important for image quality assessment and enhancement. This paper presents a method for moiré pattern detection that focuses on stability and efficiency with an evaluated loss function. The proposed approach combines frequency-domain analysis with spatial features to robustly identify moiré artifacts across different image types and interference frequencies.
Transition in focus of prediction tasks for skeleton graph component detection with Transformer

Transition Prediction tasks Skeleton graph Component detection Transformer action recognition graph neural network

Zhiyuan Wang , Cong Yang , Yulu Zhang , Zeyd Boukhers , Wei Sui , Yi Ji , Chunping Liu

ACM Multimedia Asia 1-7 2024

PDF DOI
Wang, Z., Yang, C., Zhang, Y., Boukhers, Z., Sui, W., Ji, Y., & Liu, C. (2024). Transition in focus of prediction tasks for skeleton graph component detection with Transformer. ACM Multimedia Asia, 1–7. https://doi.org/10.1145/3696409.3700170

Wang, Zhiyuan, , and Cong Yang, , and Yulu Zhang, , and Zeyd Boukhers, , and Wei Sui, , and Yi Ji, , and Chunping Liu. "Transition in focus of prediction tasks for skeleton graph component detection with Transformer." ACM Multimedia Asia, 2024, pp. 1-7.

Wang, Zhiyuan, , and Cong Yang, , and Yulu Zhang, , and Zeyd Boukhers, , and Wei Sui, , and Yi Ji, , and Chunping Liu. "Transition in focus of prediction tasks for skeleton graph component detection with Transformer." ACM Multimedia Asia (2024): 1-7.
@inproceedings{Wang2024transition, title = {Transition in focus of prediction tasks for skeleton graph component detection with Transformer}, author = {Zhiyuan Wang and Cong Yang and Yulu Zhang and Zeyd Boukhers and Wei Sui and Yi Ji and Chunping Liu}, booktitle = {ACM Multimedia Asia}, pages = {1-7}, year = {2024}, doi = {10.1145/3696409.3700170}, url = {https://doi.org/10.1145/3696409.3700170}, publisher = {ACM}, }
Skeleton-based action recognition and component detection rely on graph neural networks that model the human body as a graph of joints. This paper proposes a transformer-based approach that focuses on transitions in prediction tasks for skeleton graph component detection. By attending to the transition states between different body components and action phases, our method improves the detection of skeleton graph components. Experiments on benchmark datasets demonstrate competitive performance.
Large language model in medical informatics: direct classification and enhanced text representations for automatic ICD coding

large language models ICD coding medical informatics clinical text text classification NLP healthcare AI MIMIC

Zeyd Boukhers , AmeerAli Khan , Qusai Ramadan , Cong Yang

IEEE International Conference on Bioinformatics and Biomedicine (BIBM) 1-4 2024

PDF DOI
Boukhers, Z., Khan, A. A., Ramadan, Q., & Yang, C. (2024). Large language model in medical informatics: direct classification and enhanced text representations for automatic ICD coding. IEEE International Conference on Bioinformatics and Biomedicine (BIBM), 1–4. https://doi.org/10.1109/bibm62325.2024.10822419

Boukhers, Zeyd, , and AmeerAli Khan, , and Qusai Ramadan, , and Cong Yang. "Large language model in medical informatics: direct classification and enhanced text representations for automatic ICD coding." IEEE International Conference on Bioinformatics and Biomedicine (BIBM), 2024, pp. 1-4.

Boukhers, Zeyd, , and AmeerAli Khan, , and Qusai Ramadan, , and Cong Yang. "Large language model in medical informatics: direct classification and enhanced text representations for automatic ICD coding." IEEE International Conference on Bioinformatics and Biomedicine (BIBM) (2024): 1-4.
@inproceedings{Boukhers2024large, title = {Large language model in medical informatics: direct classification and enhanced text representations for automatic ICD coding}, author = {Zeyd Boukhers and AmeerAli Khan and Qusai Ramadan and Cong Yang}, booktitle = {IEEE International Conference on Bioinformatics and Biomedicine (BIBM)}, pages = {1-4}, year = {2024}, doi = {10.1109/bibm62325.2024.10822419}, url = {https://doi.org/10.1109/bibm62325.2024.10822419}, publisher = {IEEE}, }
International Classification of Diseases (ICD) coding from clinical notes is a critical task in medical informatics that automates the assignment of diagnostic codes to patient records. Recent advances in large language models (LLMs) offer new opportunities for improving ICD coding accuracy. This paper investigates the application of LLMs in medical informatics for ICD coding, exploring two strategies: direct classification using LLMs as classifiers, and enhanced text representations by using LLMs to generate richer feature representations for downstream classification. Experiments on standard clinical benchmarks demonstrate that both strategies yield improvements over traditional deep learning approaches, with the enhanced representation strategy showing particularly strong results.
FDO manager: minimum viable FAIR digital object implementation

FAIR Digital Objects data management machine-actionability research artefacts FDO Manager reproducibility FAIR principles

Oussama Zoubia , Nagaraj Bahubali Asundi , Sezin Dogan , Adamantios Koumpis , Christoph Lange , Oya Beyan , Zeyd Boukhers

FDO Summit Berlin 2024 Proceedings 5 1-2 2024

PDF DOI
Zoubia, O., Asundi, N. B., Dogan, S., Koumpis, A., Lange, C., Beyan, O., & Boukhers, Z. (2024). FDO manager: minimum viable FAIR digital object implementation. FDO Summit Berlin 2024 Proceedings, 5, 1–2. https://doi.org/10.52825/ocp.v5i.1421

Zoubia, Oussama, , and Nagaraj Bahubali Asundi, , and Sezin Dogan, , and Adamantios Koumpis, , and Christoph Lange, , and Oya Beyan, , and Zeyd Boukhers. "FDO manager: minimum viable FAIR digital object implementation." FDO Summit Berlin 2024 Proceedings, vol. 5, 2024, pp. 1-2.

Zoubia, Oussama, , and Nagaraj Bahubali Asundi, , and Sezin Dogan, , and Adamantios Koumpis, , and Christoph Lange, , and Oya Beyan, , and Zeyd Boukhers. "FDO manager: minimum viable FAIR digital object implementation." FDO Summit Berlin 2024 Proceedings 5 (2024): 1-2.
@inproceedings{Zoubia2024fdo, title = {FDO manager: minimum viable FAIR digital object implementation}, author = {Oussama Zoubia and Nagaraj Bahubali Asundi and Sezin Dogan and Adamantios Koumpis and Christoph Lange and Oya Beyan and Zeyd Boukhers}, booktitle = {FDO Summit Berlin 2024 Proceedings}, volume = {5}, pages = {1-2}, year = {2024}, doi = {10.52825/ocp.v5i.1421}, url = {https://doi.org/10.52825/ocp.v5i.1421}, publisher = {TIB Open Publishing}, }
In the digital age, data has emerged as one of the most valuable assets across various sectors, including academia, industry, and healthcare. Effective data preservation involves the management of data to ensure its long-term accessibility and usability. Given the importance and sensitivity of data, the need for effective management is a crucial necessity. One of the big recent proposed approaches for data management is the FAIR Digital Objects (FDOs) which has emerged to revolutionize the field of data management and preservation. Central to this revolution is the alignment of FDOs with the FAIR principles (Findable, Accessible, Interoperable, Reusable), particularly emphasizing machine-actionability and interoperability across diverse data ecosystems. This paper presents "FDO Manager," a Minimum Viable Implementation of FDOs, tailored specifically for the use case and field of research artefacts such as datasets, publications, and code. The paper discusses the core ideas behind the FDO Manager, its architecture, usage and implementation details, as well as its potential impact, demonstrating a simple and abstract implementation of FDOs in the research realm.

2023

Data trading and monetization: challenges and open research directions

data trading monetization data pricing data economy data markets privacy distributed systems

Qusai Ramadan , Zeyd Boukhers , Muath AlShaikh , Christoph Lange , Jan Jürjens

The 7th International Conference on Future Networks and Distributed Systems 344-351 2023

PDF DOI
Ramadan, Q., Boukhers, Z., AlShaikh, M., Lange, C., & Jürjens, J. (2023). Data trading and monetization: challenges and open research directions. The 7th International Conference on Future Networks and Distributed Systems, 344–351. https://doi.org/10.1145/3644713.3644758

Ramadan, Qusai, , and Zeyd Boukhers, , and Muath AlShaikh, , and Christoph Lange, , and Jan Jürjens. "Data trading and monetization: challenges and open research directions." The 7th International Conference on Future Networks and Distributed Systems, 2023, pp. 344-351.

Ramadan, Qusai, , and Zeyd Boukhers, , and Muath AlShaikh, , and Christoph Lange, , and Jan Jürjens. "Data trading and monetization: challenges and open research directions." The 7th International Conference on Future Networks and Distributed Systems (2023): 344-351.
@inproceedings{Ramadan2023data, title = {Data trading and monetization: challenges and open research directions}, author = {Qusai Ramadan and Zeyd Boukhers and Muath AlShaikh and Christoph Lange and Jan Jürjens}, booktitle = {The 7th International Conference on Future Networks and Distributed Systems}, pages = {344-351}, year = {2023}, doi = {10.1145/3644713.3644758}, url = {https://doi.org/10.1145/3644713.3644758}, publisher = {ACM}, }
Data trading and monetization have become increasingly important with the rise of the data economy. This paper presents a comprehensive survey of the challenges and open research directions in data trading and monetization, covering topics such as data pricing mechanisms, market design, data valuation, privacy-preserving data trading, and the legal and ethical challenges associated with commercializing data in distributed systems.
NFDI4DS Infrastructure and Services

NFDI4DataScience research data infrastructure data science AI FAIR data data management Germany

Sonja Schimmler , Bianca Wentzel , Arnim Bleier , Stefan Dietze , Saurav Karmakar , Peter Mutschke , Angelie Kraft , Tilahun A Taffa , Ricardo Usbeck , Zeyd Boukhers , Sören Auer , Leyla J Castro , Marcel R Ackermann , Thomas Neumuth , Daniel Schneider , Ziawash Abedjan , Atif Latif , Fidan Limani , Raia Abu Ahmad , Georg Rehm , Sima Attar Khorasani , Matthias Lieber

INFORMATIK 2023 - Designing Futures: Zukünfte gestalten 2023

PDF DOI
Schimmler, S., Wentzel, B., Bleier, A., Dietze, S., Karmakar, S., Mutschke, P., Kraft, A., Taffa, T. A., Usbeck, R., Boukhers, Z., Auer, S., Castro, L. J., Ackermann, M. R., Neumuth, T., Schneider, D., Abedjan, Z., Latif, A., Limani, F., Ahmad, R. A., … Lieber, M. (2023, December 2). NFDI4DS Infrastructure and Services. INFORMATIK 2023 - Designing Futures: Zukünfte Gestalten. https://doi.org/10.18420/inf2023_103

Schimmler, Sonja, , and Bianca Wentzel, , and Arnim Bleier, , and Stefan Dietze, , and Saurav Karmakar, , and Peter Mutschke, , and Angelie Kraft, , and Tilahun A Taffa, , and Ricardo Usbeck, , and Zeyd Boukhers, , and Sören Auer, , and Leyla J Castro, , and Marcel R Ackermann, , and Thomas Neumuth, , and Daniel Schneider, , and Ziawash Abedjan, , and Atif Latif, , and Fidan Limani, , and Raia Abu Ahmad, , and Georg Rehm, , and Sima Attar Khorasani, , and Matthias Lieber. "NFDI4DS Infrastructure and Services." INFORMATIK 2023 - Designing Futures: Zukünfte gestalten, 2023.

Schimmler, Sonja, , and Bianca Wentzel, , and Arnim Bleier, , and Stefan Dietze, , and Saurav Karmakar, , and Peter Mutschke, , and Angelie Kraft, , and Tilahun A Taffa, , and Ricardo Usbeck, , and Zeyd Boukhers, , and Sören Auer, , and Leyla J Castro, , and Marcel R Ackermann, , and Thomas Neumuth, , and Daniel Schneider, , and Ziawash Abedjan, , and Atif Latif, , and Fidan Limani, , and Raia Abu Ahmad, , and Georg Rehm, , and Sima Attar Khorasani, , and Matthias Lieber. "NFDI4DS Infrastructure and Services." INFORMATIK 2023 - Designing Futures: Zukünfte gestalten (2023).
@inproceedings{Schimmler2023nfdi4ds, title = {NFDI4DS Infrastructure and Services}, author = {Sonja Schimmler and Bianca Wentzel and Arnim Bleier and Stefan Dietze and Saurav Karmakar and Peter Mutschke and Angelie Kraft and Tilahun A Taffa and Ricardo Usbeck and Zeyd Boukhers and Sören Auer and Leyla J Castro and Marcel R Ackermann and Thomas Neumuth and Daniel Schneider and Ziawash Abedjan and Atif Latif and Fidan Limani and Raia Abu Ahmad and Georg Rehm and Sima Attar Khorasani and Matthias Lieber}, booktitle = {INFORMATIK 2023 - Designing Futures: Zukünfte gestalten}, year = {2023}, doi = {10.18420/inf2023_103}, url = {https://doi.org/10.18420/inf2023_103}, }
This paper presents the infrastructure and services provided by NFDI4DataScience (NFDI4DS), a national research data infrastructure consortium in Germany dedicated to supporting data science and artificial intelligence research. The paper describes the technical components, service offerings, and integration strategies that enable researchers to manage, share, and reuse data science artifacts in compliance with FAIR principles.
Research knowledge graphs in NFDI4DS

research knowledge graphs NFDI4DataScience data science ontology research data management knowledge discovery

Saurav Karmakar , Matthäus Zloch , Fidan Limani , Benjamin Zapilko , Sharmila Upadhyaya , Jennifer D’Souza , Leyla J Castro , Georg Rehm , Marcel R Ackermann , Harald Sack , Zeyd Boukhers , Sonja Schimmler , Danilo Dessí , Peter Mutschke , Stefan Dietze

INFORMATIK 2023 - Designing Futures: Zukünfte gestalten 2023

PDF DOI
Karmakar, S., Zloch, M., Limani, F., Zapilko, B., Upadhyaya, S., D’Souza, J., Castro, L. J., Rehm, G., Ackermann, M. R., Sack, H., Boukhers, Z., Schimmler, S., Dessí, D., Mutschke, P., & Dietze, S. (2023, December 2). Research knowledge graphs in NFDI4DS. INFORMATIK 2023 - Designing Futures: Zukünfte Gestalten. https://doi.org/10.18420/inf2023_102

Karmakar, Saurav, , and Matthäus Zloch, , and Fidan Limani, , and Benjamin Zapilko, , and Sharmila Upadhyaya, , and Jennifer D’Souza, , and Leyla J Castro, , and Georg Rehm, , and Marcel R Ackermann, , and Harald Sack, , and Zeyd Boukhers, , and Sonja Schimmler, , and Danilo Dessí, , and Peter Mutschke, , and Stefan Dietze. "Research knowledge graphs in NFDI4DS." INFORMATIK 2023 - Designing Futures: Zukünfte gestalten, 2023.

Karmakar, Saurav, , and Matthäus Zloch, , and Fidan Limani, , and Benjamin Zapilko, , and Sharmila Upadhyaya, , and Jennifer D’Souza, , and Leyla J Castro, , and Georg Rehm, , and Marcel R Ackermann, , and Harald Sack, , and Zeyd Boukhers, , and Sonja Schimmler, , and Danilo Dessí, , and Peter Mutschke, , and Stefan Dietze. "Research knowledge graphs in NFDI4DS." INFORMATIK 2023 - Designing Futures: Zukünfte gestalten (2023).
@inproceedings{Karmakar2023research, title = {Research knowledge graphs in NFDI4DS}, author = {Saurav Karmakar and Matthäus Zloch and Fidan Limani and Benjamin Zapilko and Sharmila Upadhyaya and Jennifer D’Souza and Leyla J Castro and Georg Rehm and Marcel R Ackermann and Harald Sack and Zeyd Boukhers and Sonja Schimmler and Danilo Dessí and Peter Mutschke and Stefan Dietze}, booktitle = {INFORMATIK 2023 - Designing Futures: Zukünfte gestalten}, year = {2023}, doi = {10.18420/inf2023_102}, url = {https://doi.org/10.18420/inf2023_102}, }
This paper presents the research knowledge graph infrastructure developed within NFDI4DataScience (NFDI4DS). The knowledge graph connects research artifacts, datasets, publications, and researchers in the data science and AI domain, enabling discovery and reuse of research knowledge. The paper describes the ontological foundations, data integration pipelines, and use cases for the knowledge graph.
FAIR Data Spaces - Building a Common Cloud-Based Data Space for Industry and Science by Linking Gaia-X and NFDI

FAIR data spaces Gaia-X NFDI data sovereignty cloud computing data sharing research infrastructure industry-science collaboration

Zeyd Boukhers , Daniela Mockler

1st Conference on Research Data Infrastructure (CoRDI) - Connecting Communities 2023

PDF DOI
Boukhers, Z., & Mockler, D. (2023, September 1). FAIR Data Spaces - Building a Common Cloud-Based Data Space for Industry and Science by Linking Gaia-X and NFDI. 1st Conference on Research Data Infrastructure (CoRDI) - Connecting Communities. https://doi.org/10.52825/cordi.v1i.406

Boukhers, Zeyd, , and Daniela Mockler. "FAIR Data Spaces - Building a Common Cloud-Based Data Space for Industry and Science by Linking Gaia-X and NFDI." 1st Conference on Research Data Infrastructure (CoRDI) - Connecting Communities, 2023.

Boukhers, Zeyd, , and Daniela Mockler. "FAIR Data Spaces - Building a Common Cloud-Based Data Space for Industry and Science by Linking Gaia-X and NFDI." 1st Conference on Research Data Infrastructure (CoRDI) - Connecting Communities (2023).
@inproceedings{Boukhers2023fair, title = {FAIR Data Spaces - Building a Common Cloud-Based Data Space for Industry and Science by Linking Gaia-X and NFDI}, author = {Zeyd Boukhers and Daniela Mockler}, booktitle = {1st Conference on Research Data Infrastructure (CoRDI) - Connecting Communities}, year = {2023}, doi = {10.52825/cordi.v1i.406}, url = {https://zeyd.boukhers.com/wp-content/uploads/2023/12/FAIR-Data-Spaces-Posterabstract-CoRDI-2023.pdf}, }
This paper presents the concept of FAIR Data Spaces as a common cloud-based data space that bridges industry and science by linking Gaia-X and NFDI (National Research Data Infrastructure). The paper discusses how FAIR principles can be applied to build interoperable data spaces that enable data sovereignty while facilitating data sharing across domains. It presents the architecture and implementation of a prototype that connects Gaia-X data spaces with NFDI research data infrastructure.
Beyond trading data: the hidden influence of public awareness and interest on cryptocurrency volatility

cryptocurrency volatility prediction Bitcoin social media public sentiment machine learning LSTM AdaBoost multimodal model trading data

Zeyd Boukhers , Azeddine Bouabdallah , Cong Yang , Jan Jürjens

32nd ACM International Conference on Information and Knowledge Management (CIKM) 142-151 2023

PDF DOI
Boukhers, Z., Bouabdallah, A., Yang, C., & Jürjens, J. (2023). Beyond trading data: the hidden influence of public awareness and interest on cryptocurrency volatility. 32nd ACM International Conference on Information and Knowledge Management (CIKM), 142–151. https://doi.org/10.1145/3583780.3614790

Boukhers, Zeyd, , and Azeddine Bouabdallah, , and Cong Yang, , and Jan Jürjens. "Beyond trading data: the hidden influence of public awareness and interest on cryptocurrency volatility." 32nd ACM International Conference on Information and Knowledge Management (CIKM), 2023, pp. 142-151.

Boukhers, Zeyd, , and Azeddine Bouabdallah, , and Cong Yang, , and Jan Jürjens. "Beyond trading data: the hidden influence of public awareness and interest on cryptocurrency volatility." 32nd ACM International Conference on Information and Knowledge Management (CIKM) (2023): 142-151.
@inproceedings{Bukhers2023beyond, title = {Beyond trading data: the hidden influence of public awareness and interest on cryptocurrency volatility}, author = {Zeyd Boukhers and Azeddine Bouabdallah and Cong Yang and Jan Jürjens}, booktitle = {32nd ACM International Conference on Information and Knowledge Management (CIKM)}, pages = {142-151}, year = {2023}, doi = {10.1145/3583780.3614790}, url = {https://doi.org/10.1145/3583780.3614790}, publisher = {ACM}, }
Cryptocurrencies have emerged as a new class of decentralized financial assets since the introduction of Bitcoin in 2009. The volatile nature of cryptocurrency markets has attracted significant attention from researchers and investors. This paper investigates the factors that drive cryptocurrency volatility beyond traditional trading data. We propose CoMForE, a multimodal AdaBoost-LSTM ensemble model that incorporates not only historical trading data but also public sentiments extracted from related tweets, public interest quantified through search volumes, and blockchain hash-rate data. Our experiments on Bitcoin demonstrate that integrating these non-trading factors significantly improves volatility prediction performance, highlighting the hidden influence of public awareness and interest on cryptocurrency markets.
Enhancing Reproducibility in Research through FAIR Digital Objects

FAIR principles FAIR Digital Objects reproducibility research artifacts data management NFDI4DataScience

Zeyd Boukhers , Leyla Jael Garcia Castro

1st Conference on Research Data Infrastructure 1 2023

PDF DOI
Boukhers, Z., & Castro, L. J. G. (2023). Enhancing Reproducibility in Research through FAIR Digital Objects. 1st Conference on Research Data Infrastructure, 1. https://doi.org/10.52825/cordi.v1i.406

Boukhers, Zeyd, , and Leyla Jael Garcia Castro. "Enhancing Reproducibility in Research through FAIR Digital Objects." 1st Conference on Research Data Infrastructure, vol. 1, 2023.

Boukhers, Zeyd, , and Leyla Jael Garcia Castro. "Enhancing Reproducibility in Research through FAIR Digital Objects." 1st Conference on Research Data Infrastructure 1 (2023).
@inproceedings{Boukhers2023enhancing, title = {Enhancing Reproducibility in Research through FAIR Digital Objects}, author = {Zeyd Boukhers and Leyla Jael Garcia Castro}, booktitle = {1st Conference on Research Data Infrastructure}, volume = {1}, year = {2023}, doi = {10.52825/cordi.v1i.406}, url = {https://doi.org/10.52825/cordi.v1i.406}, publisher = {TIB Open Publishing}, }
The FAIR principles were introduced to enhance data reuse by providing guidelines for effective data management practices. In the broader context of research, assets encompass not only data but also artifacts such as code, software, and publications. FAIRifying these artifacts is as essential as FAIRifying data, given the increasing complexity of current AI approaches that make reproducibility extremely challenging. Therefore, the reuse of these artifacts is growing in importance. The concept of FAIR Digital Objects (FDOs) presents a solution to FAIRify these artifacts, treating them as FDOs. NFDI4DataScience is embracing FDOs and proposing an architecture to efficiently manage them.
PADME-SoSci: A Platform for Analytics and Distributed Machine Learning for the Social Sciences

distributed machine learning social sciences PADME privacy-preserving analytics federated learning Personal Health Train data sovereignty

Zeyd Boukhers , Arnim Bleier , Yeliz Üçer Yediel , Mio Hienstorfer-Heitmann , Mehrshad Jaberansary , Adamantios Koumpis , Oya Beyan

ACM/IEEE JCDL 2023 - Joint Conference on Digital Libraries 251-252 2023

PDF DOI
Boukhers, Z., Bleier, A., Yediel, Y. Ü., Hienstorfer-Heitmann, M., Jaberansary, M., Koumpis, A., & Beyan, O. (2023). PADME-SoSci: A Platform for Analytics and Distributed Machine Learning for the Social Sciences. ACM/IEEE JCDL 2023 - Joint Conference on Digital Libraries, 251–252. https://doi.org/10.1109/jcdl57899.2023.00047

Boukhers, Zeyd, , and Arnim Bleier, , and Yeliz Üçer Yediel, , and Mio Hienstorfer-Heitmann, , and Mehrshad Jaberansary, , and Adamantios Koumpis, , and Oya Beyan. "PADME-SoSci: A Platform for Analytics and Distributed Machine Learning for the Social Sciences." ACM/IEEE JCDL 2023 - Joint Conference on Digital Libraries, 2023, pp. 251-252.

Boukhers, Zeyd, , and Arnim Bleier, , and Yeliz Üçer Yediel, , and Mio Hienstorfer-Heitmann, , and Mehrshad Jaberansary, , and Adamantios Koumpis, , and Oya Beyan. "PADME-SoSci: A Platform for Analytics and Distributed Machine Learning for the Social Sciences." ACM/IEEE JCDL 2023 - Joint Conference on Digital Libraries (2023): 251-252.
@inproceedings{Boukhers2023padme, title = {PADME-SoSci: A Platform for Analytics and Distributed Machine Learning for the Social Sciences}, author = {Zeyd Boukhers and Arnim Bleier and Yeliz Üçer Yediel and Mio Hienstorfer-Heitmann and Mehrshad Jaberansary and Adamantios Koumpis and Oya Beyan}, booktitle = {ACM/IEEE JCDL 2023 - Joint Conference on Digital Libraries}, pages = {251-252}, year = {2023}, doi = {10.1109/jcdl57899.2023.00047}, url = {https://doi.org/10.1109/jcdl57899.2023.00047}, publisher = {IEEE}, }
Social science research increasingly requires complex data analytics and machine learning capabilities. This paper introduces PADME-SoSci, a platform for analytics and distributed machine learning tailored for the social sciences. PADME-SoSci enables researchers to apply machine learning models to sensitive social science data without transferring the data from its original location, using a federated learning approach based on the Personal Health Train (PHT) paradigm. The platform supports privacy-preserving analytics through a data space architecture that respects data sovereignty.
Enhancing data space semantic interoperability through Machine Learning: a visionary perspective

data spaces semantic interoperability machine learning ontology alignment FAIR data data sovereignty Gaia-X federated data ecosystems

Zeyd Boukhers , Oya Beyan , Christoph Lange

Companion Proceedings of the ACM Web Conference 2023 (WWW ’23 Companion) 1462-1467 2023

PDF DOI
Boukhers, Z., Beyan, O., & Lange, C. (2023). Enhancing data space semantic interoperability through Machine Learning: a visionary perspective. Companion Proceedings of the ACM Web Conference 2023 (WWW ’23 Companion), 1462–1467. https://doi.org/10.1145/3543873.3587658

Boukhers, Zeyd, , and Oya Beyan, , and Christoph Lange. "Enhancing data space semantic interoperability through Machine Learning: a visionary perspective." Companion Proceedings of the ACM Web Conference 2023 (WWW ’23 Companion), 2023, pp. 1462-1467.

Boukhers, Zeyd, , and Oya Beyan, , and Christoph Lange. "Enhancing data space semantic interoperability through Machine Learning: a visionary perspective." Companion Proceedings of the ACM Web Conference 2023 (WWW ’23 Companion) (2023): 1462-1467.
@inproceedings{Boukhers2023enhancingdata, title = {Enhancing data space semantic interoperability through Machine Learning: a visionary perspective}, author = {Zeyd Boukhers and Oya Beyan and Christoph Lange}, booktitle = {Companion Proceedings of the ACM Web Conference 2023 (WWW ’23 Companion)}, pages = {1462-1467}, year = {2023}, doi = {10.1145/3543873.3587658}, url = {https://doi.org/10.1145/3543873.3587658}, publisher = {ACM}, }
Data spaces are emerging as a key infrastructure for enabling data sharing and sovereignty in the digital economy. However, ensuring semantic interoperability across heterogeneous data spaces remains a challenge. This paper presents a visionary perspective on enhancing data space semantic interoperability through machine learning. We propose a framework that leverages ML techniques to automatically align and translate ontologies, schemas, and data models across different data spaces, reducing the manual effort required for semantic integration. The paper discusses the technical challenges and potential solutions for achieving seamless semantic interoperability in federated data ecosystems.

2022

Whois? Deep author name disambiguation using bibliographic data

author name disambiguation DBLP co-author network bibliographic data digital libraries name ambiguity deep learning

Zeyd Boukhers , Nagaraj Bahubali

The 26th International Conference on Theory and Practice of Digital Libraries 201-215 2022

PDF DOI
Boukhers, Z., & Bahubali, N. (2022). Whois? Deep author name disambiguation using bibliographic data. The 26th International Conference on Theory and Practice of Digital Libraries, 201–215. https://doi.org/10.1007/978-3-031-16802-4_16

Boukhers, Zeyd, , and Nagaraj Bahubali. "Whois? Deep author name disambiguation using bibliographic data." The 26th International Conference on Theory and Practice of Digital Libraries, 2022, pp. 201-215.

Boukhers, Zeyd, , and Nagaraj Bahubali. "Whois? Deep author name disambiguation using bibliographic data." The 26th International Conference on Theory and Practice of Digital Libraries (2022): 201-215.
@inproceedings{Boukhers2022whois, title = {Whois? Deep author name disambiguation using bibliographic data}, author = {Zeyd Boukhers and Nagaraj Bahubali}, booktitle = {The 26th International Conference on Theory and Practice of Digital Libraries}, pages = {201-215}, year = {2022}, doi = {10.1007/978-3-031-16802-4_16}, url = {https://doi.org/10.1007/978-3-031-16802-4_16}, publisher = {Springer}, }
As the number of authors is increasing exponentially over the years, the number of authors sharing the same names is increasing proportionally. This makes it challenging to assign newly published papers to their adequate authors. Therefore, Author Name Ambiguity (ANA) is considered a critical open problem in digital libraries. This paper proposes an Author Name Disambiguation (AND) approach that links author names to their real-world entities by leveraging their co-authors and domain of research. The approach uses a collection from the DBLP repository that contains more than 5 million bibliographic records authored by around 2.6 million co-authors. It first groups authors who share the same last names and same first name initials. The author within each group is identified by capturing the relation with his/her co-authors and area of research, represented by the titles of the validated publications of the corresponding author. This paper won the Best Paper Award at TPDL 2022.
Vision and Natural Language for Metadata Extraction from PDF Documents: A Multimodal Approach

metadata extraction PDF documents multimodal learning computer vision natural language processing scientific publications digital libraries

Zeyd Boukhers , Azeddine Bouabdallah

ACM/IEEE Joint Conference on Digital Library 1-5 2022

PDF DOI
Boukhers, Z., & Bouabdallah, A. (2022). Vision and Natural Language for Metadata Extraction from PDF Documents: A Multimodal Approach. ACM/IEEE Joint Conference on Digital Library, 1–5. https://doi.org/10.1145/3529372.3533295

Boukhers, Zeyd, , and Azeddine Bouabdallah. "Vision and Natural Language for Metadata Extraction from PDF Documents: A Multimodal Approach." ACM/IEEE Joint Conference on Digital Library, 2022, pp. 1-5.

Boukhers, Zeyd, , and Azeddine Bouabdallah. "Vision and Natural Language for Metadata Extraction from PDF Documents: A Multimodal Approach." ACM/IEEE Joint Conference on Digital Library (2022): 1-5.
@inproceedings{Boukhers2022vision, title = {Vision and Natural Language for Metadata Extraction from PDF Documents: A Multimodal Approach}, author = {Zeyd Boukhers and Azeddine Bouabdallah}, booktitle = {ACM/IEEE Joint Conference on Digital Library}, pages = {1-5}, year = {2022}, doi = {10.1145/3529372.3533295}, url = {https://doi.org/10.1145/3529372.3533295}, publisher = {ACM}, }
Extracting metadata from scientific PDF documents is a key challenge for building digital libraries and research knowledge graphs. Existing approaches rely either on NLP or computer vision techniques independently. In this paper, we propose a multimodal neural network approach that combines both vision and natural language processing for metadata extraction from scientific PDF documents. Our approach processes the document both as an image (capturing layout and visual cues) and as text (capturing semantic content), and fuses these modalities to achieve higher accuracy. Experiments on benchmark datasets demonstrate that our multimodal approach significantly outperforms unimodal baselines.

2021

MexPub: Deep transfer learning for metadata extraction from German publications

metadata extraction German publications deep learning PDF Mask R-CNN scientific documents document layout analysis

Zeyd Boukhers , Prantik Goswami , Nada Beili , Muhammad Arslan Zafar , Timo Hartmann

ACM/IEEE Joint Conference on Digital Library 1-8 2021

PDF DOI
Boukhers, Z., Goswami, P., Beili, N., Zafar, M. A., & Hartmann, T. (2021). MexPub: Deep transfer learning for metadata extraction from German publications. ACM/IEEE Joint Conference on Digital Library, 1–8. https://doi.org/10.1109/jcdl52503.2021.00076

Boukhers, Zeyd, , and Prantik Goswami, , and Nada Beili, , and Muhammad Arslan Zafar, , and Timo Hartmann. "MexPub: Deep transfer learning for metadata extraction from German publications." ACM/IEEE Joint Conference on Digital Library, 2021, pp. 1-8.

Boukhers, Zeyd, , and Prantik Goswami, , and Nada Beili, , and Muhammad Arslan Zafar, , and Timo Hartmann. "MexPub: Deep transfer learning for metadata extraction from German publications." ACM/IEEE Joint Conference on Digital Library (2021): 1-8.
@inproceedings{Boukhers2021mexpub, title = {MexPub: Deep transfer learning for metadata extraction from German publications}, author = {Zeyd Boukhers and Prantik Goswami and Nada Beili and Muhammad Arslan Zafar and Timo Hartmann}, booktitle = {ACM/IEEE Joint Conference on Digital Library}, pages = {1-8}, year = {2021}, doi = {10.1109/jcdl52503.2021.00076}, url = {https://doi.org/10.1109/jcdl52503.2021.00076}, publisher = {IEEE}, }
Extracting metadata from scientific papers is considered a solved problem in NLP due to high-accuracy state-of-the-art methods. However, this does not apply to German scientific publications, which have a variety of styles and layouts. In contrast to most English scientific publications that follow standard and simple layouts, the order, content, position and size of metadata in German publications vary greatly among publications. This variety makes traditional NLP methods fail to accurately extract metadata. We present MexPub, a method that extracts metadata from PDF documents with different layouts and styles by viewing the document as an image. Our method uses Mask R-CNN trained on the COCO dataset and fine-tuned with PubLayNet and a proposed synthetic dataset of 30K article snapshots to extract nine metadata patterns including author and title. Experiments show that MexPub achieves high accuracy on German scientific publications.
BibliODAP’21: The 1st workshop on bibliographic data analysis and processing

bibliographic data digital libraries scientometrics reference extraction metadata citation analysis workshop

Zeyd Boukhers , Philipp Mayr , Silvio Peroni

Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining 4110-4111 2021

PDF DOI
Boukhers, Z., Mayr, P., & Peroni, S. (2021). BibliODAP’21: The 1st workshop on bibliographic data analysis and processing. Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 4110–4111. https://doi.org/10.1145/3447548.3469482

Boukhers, Zeyd, , and Philipp Mayr, , and Silvio Peroni. "BibliODAP’21: The 1st workshop on bibliographic data analysis and processing." Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2021, pp. 4110-4111.

Boukhers, Zeyd, , and Philipp Mayr, , and Silvio Peroni. "BibliODAP’21: The 1st workshop on bibliographic data analysis and processing." Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (2021): 4110-4111.
@inproceedings{Boukhers2021bibliodap, title = {BibliODAP’21: The 1st workshop on bibliographic data analysis and processing}, author = {Zeyd Boukhers and Philipp Mayr and Silvio Peroni}, booktitle = {Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery and Data Mining}, pages = {4110-4111}, year = {2021}, doi = {10.1145/3447548.3469482}, url = {https://doi.org/10.1145/3447548.3469482}, publisher = {ACM}, }
Bibliographic data analysis and processing has become increasingly important with the growth of digital libraries and the need for scientometrics research. This paper presents an overview of the BibliODAP’21 workshop, the first workshop on bibliographic data analysis and processing co-located with ACM SIGKDD 2021. The workshop brought together researchers working on reference extraction, metadata quality assessment, author disambiguation, citation analysis, and other bibliographic data processing tasks.

2019

An end-to-end approach for extracting and segmenting high-variance references from PDF documents

reference extraction PDF documents end-to-end learning citation segmentation probabilistic approach digital libraries scientific publications

Zeyd Boukhers , Shrihash Ambhore , Steffen Staab

ACM/IEEE Joint Conference on Digital Library 1-10 2019

PDF DOI
Boukhers, Z., Ambhore, S., & Staab, S. (2019). An end-to-end approach for extracting and segmenting high-variance references from PDF documents. ACM/IEEE Joint Conference on Digital Library, 1–10. https://doi.org/10.1109/jcdl.2019.00035

Boukhers, Zeyd, , and Shrihash Ambhore, , and Steffen Staab. "An end-to-end approach for extracting and segmenting high-variance references from PDF documents." ACM/IEEE Joint Conference on Digital Library, 2019, pp. 1-10.

Boukhers, Zeyd, , and Shrihash Ambhore, , and Steffen Staab. "An end-to-end approach for extracting and segmenting high-variance references from PDF documents." ACM/IEEE Joint Conference on Digital Library (2019): 1-10.
@inproceedings{Boukhers2019an, title = {An end-to-end approach for extracting and segmenting high-variance references from PDF documents}, author = {Zeyd Boukhers and Shrihash Ambhore and Steffen Staab}, booktitle = {ACM/IEEE Joint Conference on Digital Library}, pages = {1-10}, year = {2019}, doi = {10.1109/jcdl.2019.00035}, url = {https://doi.org/10.1109/jcdl.2019.00035}, publisher = {ACM}, }
References extraction from PDF documents is a well-studied problem, but existing methods fail on documents with high-variance references—references that vary greatly in content, length, and location within the document. This paper proposes an end-to-end approach for extracting and segmenting high-variance references from PDF documents. Unlike the classical sequential pipeline, our probabilistic approach learns the different characteristics of references in a coherent scheme that reduces error accumulation. The method achieves state-of-the-art performance on challenging benchmarks containing documents from diverse scientific domains.
EXCITE – A toolchain to extract, match and publish open literature references

reference extraction EXCITE toolchain open citations PDF scientific publications bibliometrics citation matching

Azam Hosseini , Behnam Ghavimi , Zeyd Boukhers , Philipp Mayr

ACM/IEEE Joint Conference on Digital Library 432-433 2019

PDF DOI
Hosseini, A., Ghavimi, B., Boukhers, Z., & Mayr, P. (2019). EXCITE – A toolchain to extract, match and publish open literature references. ACM/IEEE Joint Conference on Digital Library, 432–433. https://doi.org/10.1109/jcdl.2019.00105

Hosseini, Azam, , and Behnam Ghavimi, , and Zeyd Boukhers, , and Philipp Mayr. "EXCITE – A toolchain to extract, match and publish open literature references." ACM/IEEE Joint Conference on Digital Library, 2019, pp. 432-433.

Hosseini, Azam, , and Behnam Ghavimi, , and Zeyd Boukhers, , and Philipp Mayr. "EXCITE – A toolchain to extract, match and publish open literature references." ACM/IEEE Joint Conference on Digital Library (2019): 432-433.
@inproceedings{Hossein i2019excite, title = {EXCITE – A toolchain to extract, match and publish open literature references}, author = {Azam Hosseini and Behnam Ghavimi and Zeyd Boukhers and Philipp Mayr}, booktitle = {ACM/IEEE Joint Conference on Digital Library}, pages = {432-433}, year = {2019}, doi = {10.1109/jcdl.2019.00105}, url = {https://doi.org/10.1109/jcdl.2019.00105}, publisher = {ACM}, }
Open citations are critical for the advancement of bibliometrics and scientometrics research. This paper presents EXCITE, a generic toolchain to extract, segment and match literature references from full-text PDF files. EXCITE is designed to extract citations from social science publications and make more citation data publicly available. The toolchain integrates components for PDF text extraction, reference string detection, reference segmentation, and reference matching against databases. A public demo system integrates all pipeline components under a user-friendly interface, and a special component ingests extracted references into the Open Citation Corpus.

2016

Automatic recognition of movement patterns in the Vojta therapy using RGB-D data

Vojta therapy movement pattern recognition RGB-D activity recognition physiotherapy depth sensor skeletal data cerebral palsy

Muhammad Hassan Khan , Jullien Helsp er , Zeyd Boukhers , Marcin Grzegorzek

Image Processing (ICIP), IEEE International Conference on 1235-1239 2016

PDF DOI
Khan, M. H., er, J. H., Boukhers, Z., & Grzegorzek, M. (2016). Automatic recognition of movement patterns in the Vojta therapy using RGB-D data. Image Processing (ICIP), IEEE International Conference On, 1235–1239. https://doi.org/10.1109/icip.2016.7532555

Khan, Muhammad Hassan, , and Jullien Helsp er, , and Zeyd Boukhers, , and Marcin Grzegorzek. "Automatic recognition of movement patterns in the Vojta therapy using RGB-D data." Image Processing (ICIP), IEEE International Conference on, 2016, pp. 1235-1239.

Khan, Muhammad Hassan, , and Jullien Helsp er, , and Zeyd Boukhers, , and Marcin Grzegorzek. "Automatic recognition of movement patterns in the Vojta therapy using RGB-D data." Image Processing (ICIP), IEEE International Conference on (2016): 1235-1239.
@inproceedings{Khan2016automatic, title = {Automatic recognition of movement patterns in the Vojta therapy using RGB-D data}, author = {Muhammad Hassan Khan and Jullien Helsp er and Zeyd Boukhers and Marcin Grzegorzek}, booktitle = {Image Processing (ICIP), IEEE International Conference on}, pages = {1235-1239}, year = {2016}, doi = {10.1109/icip.2016.7532555}, url = {https://doi.org/10.1109/icip.2016.7532555}, publisher = {IEEE}, }
The Vojta therapy is a neuro-developmental treatment approach used in physiotherapy for children with cerebral palsy and other neurological conditions. Manual assessment of patients’ movement patterns during therapy is time-consuming and subjective. This paper proposes an automatic approach for recognizing movement patterns in Vojta therapy using RGB-D data. We develop a method that leverages depth maps and skeletal information captured by an RGB-D sensor to automatically classify different Vojta therapy reflex positions, enabling objective and efficient therapy assessment.

2015

Shape-Based Eye Blinking Detection and Analysis

eye blinking detection shape features human-computer interaction fatigue detection pattern recognition face analysis

Zeyd Boukhers , Tomasz Jarzyński , Florian Schmidt , Oliver Tiebe , Marcin Grzegorzek

Computer Recognition Systems (CORES), International Conference on 327-335 2015

PDF DOI
Boukhers, Z., Jarzyński, T., Schmidt, F., Tiebe, O., & Grzegorzek, M. (2015). Shape-Based Eye Blinking Detection and Analysis. Computer Recognition Systems (CORES), International Conference On, 327–335. https://doi.org/10.1007/978-3-319-26227-7_31

Boukhers, Zeyd, , and Tomasz Jarzyński, , and Florian Schmidt, , and Oliver Tiebe, , and Marcin Grzegorzek. "Shape-Based Eye Blinking Detection and Analysis." Computer Recognition Systems (CORES), International Conference on, 2015, pp. 327-335.

Boukhers, Zeyd, , and Tomasz Jarzyński, , and Florian Schmidt, , and Oliver Tiebe, , and Marcin Grzegorzek. "Shape-Based Eye Blinking Detection and Analysis." Computer Recognition Systems (CORES), International Conference on (2015): 327-335.
@inproceedings{Boukhers2015shape, title = {Shape-Based Eye Blinking Detection and Analysis}, author = {Zeyd Boukhers and Tomasz Jarzyński and Florian Schmidt and Oliver Tiebe and Marcin Grzegorzek}, booktitle = {Computer Recognition Systems (CORES), International Conference on}, pages = {327-335}, year = {2015}, doi = {10.1007/978-3-319-26227-7_31}, url = {https://doi.org/10.1007/978-3-319-26227-7_31}, publisher = {Springer International Publishing}, }
Eye blinking is a natural and unconscious action that carries important information for human-computer interaction, fatigue detection, and medical diagnosis. This paper presents a shape-based method for eye blinking detection and analysis. Our approach extracts geometric features from the eye region and employs shape descriptors to characterize and classify blinking patterns. The method achieves robust detection under varying illumination conditions and head poses, making it suitable for real-world applications.
Environmental microbiological content-based image retrieval system using internal structure histogram

content-based image retrieval microbiology microscopy image descriptors histogram environmental monitoring microorganisms

Yan Ling Zou , Chen Li , Zeyd Boukhers , Kimiaki Shirahama , Tao Jiang , Marcin Grzegorzek

Computer Recognition Systems (CORES), International Conference on 543-552 2015

PDF DOI
Zou, Y. L., Li, C., Boukhers, Z., Shirahama, K., Jiang, T., & Grzegorzek, M. (2015). Environmental microbiological content-based image retrieval system using internal structure histogram. Computer Recognition Systems (CORES), International Conference On, 543–552. https://doi.org/10.1007/978-3-319-26227-7_51

Zou, Yan Ling, , and Chen Li, , and Zeyd Boukhers, , and Kimiaki Shirahama, , and Tao Jiang, , and Marcin Grzegorzek. "Environmental microbiological content-based image retrieval system using internal structure histogram." Computer Recognition Systems (CORES), International Conference on, 2015, pp. 543-552.

Zou, Yan Ling, , and Chen Li, , and Zeyd Boukhers, , and Kimiaki Shirahama, , and Tao Jiang, , and Marcin Grzegorzek. "Environmental microbiological content-based image retrieval system using internal structure histogram." Computer Recognition Systems (CORES), International Conference on (2015): 543-552.
@inproceedings{Ling2015environmental, title = {Environmental microbiological content-based image retrieval system using internal structure histogram}, author = {Yan Ling Zou and Chen Li and Zeyd Boukhers and Kimiaki Shirahama and Tao Jiang and Marcin Grzegorzek}, booktitle = {Computer Recognition Systems (CORES), International Conference on}, pages = {543-552}, year = {2015}, doi = {10.1007/978-3-319-26227-7_51}, url = {https://doi.org/10.1007/978-3-319-26227-7_51}, publisher = {Springer International Publishing}, }
This paper presents an environmental microbiological content-based image retrieval system using internal structure histogram. The system is designed to retrieve microbiological images based on visual similarity, enabling researchers to find similar microscopic specimens efficiently. We propose an internal structure histogram descriptor that captures the structural characteristics of microorganism images, and evaluate it on a benchmark dataset of environmental microbiological images.
Extracting 3D Trajectories of Objects from 2D Videos Using Particle Filter

3D trajectory extraction particle filter object detection depth estimation monocular video activity analysis multimedia retrieval

Zeyd Boukhers , Kimiaki Shirahama , Frédéric Li , Marcin Grzegorzek

Multimedia Retrieval, ACM International Conference on 83-90 2015

PDF DOI
Boukhers, Z., Shirahama, K., Li, F., & Grzegorzek, M. (2015). Extracting 3D Trajectories of Objects from 2D Videos Using Particle Filter. Multimedia Retrieval, ACM International Conference On, 83–90. https://doi.org/10.1145/2671188.2749331

Boukhers, Zeyd, , and Kimiaki Shirahama, , and Frédéric Li, , and Marcin Grzegorzek. "Extracting 3D Trajectories of Objects from 2D Videos Using Particle Filter." Multimedia Retrieval, ACM International Conference on, 2015, pp. 83-90.

Boukhers, Zeyd, , and Kimiaki Shirahama, , and Frédéric Li, , and Marcin Grzegorzek. "Extracting 3D Trajectories of Objects from 2D Videos Using Particle Filter." Multimedia Retrieval, ACM International Conference on (2015): 83-90.
@inproceedings{Boukhers2015exrtracting, title = {Extracting 3D Trajectories of Objects from 2D Videos Using Particle Filter}, author = {Zeyd Boukhers and Kimiaki Shirahama and Frédéric Li and Marcin Grzegorzek}, booktitle = {Multimedia Retrieval, ACM International Conference on}, pages = {83-90}, year = {2015}, doi = {10.1145/2671188.2749331}, url = {https://doi.org/10.1145/2671188.2749331}, publisher = {ACM}, }
Extracting 3D trajectories of objects from 2D videos is essential for understanding spatial interactions and enabling event detection. This paper introduces a method that extracts 3D trajectories by combining object detection with depth estimation using a particle filter framework. The major challenge is the inconsistency between object detection and depth estimation results. We address this by initializing the 3D position of an object using the frame with the highest consistency between detection and depth estimation, then tracking the object in 3D using particle filter. Experiments demonstrate the effectiveness of our method for trajectory extraction from monocular video.

Workshop Papers

2025

MESD: Metadata Extraction from Scholarly Documents - A Shared Task Overview

metadata extraction scholarly documents knowledge graphs shared task

Zeyd Boukhers , Cong Yang

ESWC International Workshop on Natural Scientific Language Processing and Research Knowledge Graphs (NSLP 2025) 2025

PDF DOI
Boukhers, Z., & Yang, C. (2025, April). MESD: Metadata Extraction from Scholarly Documents - A Shared Task Overview. ESWC International Workshop on Natural Scientific Language Processing and Research Knowledge Graphs (NSLP 2025). https://doi.org/00000

Boukhers, Zeyd, , and Cong Yang. "MESD: Metadata Extraction from Scholarly Documents - A Shared Task Overview." ESWC International Workshop on Natural Scientific Language Processing and Research Knowledge Graphs (NSLP 2025), 2025.

Boukhers, Zeyd, , and Cong Yang. "MESD: Metadata Extraction from Scholarly Documents - A Shared Task Overview." ESWC International Workshop on Natural Scientific Language Processing and Research Knowledge Graphs (NSLP 2025) (2025).
@inproceedings{Boukhers2025MESD, title = {MESD: Metadata Extraction from Scholarly Documents - A Shared Task Overview}, author = {Zeyd Boukhers and Cong Yang}, booktitle = {ESWC International Workshop on Natural Scientific Language Processing and Research Knowledge Graphs (NSLP 2025)}, year = {2025}, doi = {00000}, url = {https://ceur-ws.org/Vol-3977/NSLP-06.pdf}, publisher = {ESWC Workshop Proceedings}, }
This paper presents an overview of the Metadata Extraction from Scholarly Documents (MESD) shared task, which was designed to address the challenge of extracting structured metadata (e.g. Title, Author, Abstract, etc.) from scientific publications. The task aimed to promote the development of techniques for making scholarly data more Findable, Accessible, Interoperable, Reusable (FAIR) by improving metadata extraction from PDF documents. We describe the task design and the creation of two complementary datasets: (1) the S2ORC_Exp500v1 dataset consisting of 500 training samples, 100 validation samples, and 100 test samples with text-based annotations, and (2) the SSOAR Multidisciplinary Vision Dataset (SSOARGMVD) containing more than 8000 documents with bounding box annotations suitable for computer vision approaches. We discuss potential directions for future research in metadata extraction from scholarly documents, highlighting the opportunities presented by these new resources.
Towards Systematic Specification and Verification of Fairness Requirements: A Position Paper

fairness requirements specification verification software engineering ethics

Qusai Ramadan , Jukka Ruohonen , Abhishek Tiwari , Adam Alami , Zeyd Boukhers

2025 IEEE 33rd International Requirements Engineering Conference Workshops 2025

PDF DOI
Ramadan, Q., Ruohonen, J., Tiwari, A., Alami, A., & Boukhers, Z. (2025, September). Towards Systematic Specification and Verification of Fairness Requirements: A Position Paper. 2025 IEEE 33rd International Requirements Engineering Conference Workshops. https://doi.org/10.1109/rew66121.2025.00059

Ramadan, Qusai, , and Jukka Ruohonen, , and Abhishek Tiwari, , and Adam Alami, , and Zeyd Boukhers. "Towards Systematic Specification and Verification of Fairness Requirements: A Position Paper." 2025 IEEE 33rd International Requirements Engineering Conference Workshops, 2025.

Ramadan, Qusai, , and Jukka Ruohonen, , and Abhishek Tiwari, , and Adam Alami, , and Zeyd Boukhers. "Towards Systematic Specification and Verification of Fairness Requirements: A Position Paper." 2025 IEEE 33rd International Requirements Engineering Conference Workshops (2025).
@inproceedings{Ramadan2025Towards, title = {Towards Systematic Specification and Verification of Fairness Requirements: A Position Paper}, author = {Qusai Ramadan and Jukka Ruohonen and Abhishek Tiwari and Adam Alami and Zeyd Boukhers}, booktitle = {2025 IEEE 33rd International Requirements Engineering Conference Workshops}, year = {2025}, doi = {10.1109/rew66121.2025.00059}, url = {https://doi.org/10.1109/rew66121.2025.00059}, publisher = {IEEE}, }
Decisions suggested by improperly designed software systems might be prone to discriminate against people based on protected characteristics, such as gender and ethnicity. Previous studies attribute such undesired behavior to flaws in algorithmic design or biased data. However, these studies ignore that discrimination is often the result of a lack of well-specified fairness requirements and their verification. The fact that experts’ knowledge about fairness is often implicit makes the task of specifying precise and verifiable fairness requirements difficult. In related domains, such as security engineering, knowledge graphs have been proven to be effective in formalizing knowledge to assist requirements specification and verification. To address the lack of formal mechanisms for specifying and verifying fairness requirements, we propose the development of a knowledge graph-based framework for fairness. In this paper, we discuss the challenges, research questions, and a road map towards addressing the research questions.
A survey on metadata for Machine Learning models and datasets: standards, practices, and harmonization challenges

metadata machine learning models datasets standards harmonization survey

Asefa Genet , Chen Zongxiong , Oussama Zoubia , Fidan Limani , Kanishka Silva , Muhammad Asif Suryani , Benjamin Zapilko , Leyla Jael Castro , Ekaterina Kutafina , Dhwani Solanki , Heike Fliegl , Sonja Schimmler , Zeyd Boukhers , Harald Sack

Sci-K Workshop at the International Semantic Web Conference (ISWC 2025) 2025

PDF DOI
Genet, A., Zongxiong, C., Zoubia, O., Limani, F., Silva, K., Suryani, M. A., Zapilko, B., Castro, L. J., Kutafina, E., Solanki, D., Fliegl, H., Schimmler, S., Boukhers, Z., & Sack, H. (2025, September). A survey on metadata for Machine Learning models and datasets: standards, practices, and harmonization challenges. Sci-K Workshop at the International Semantic Web Conference (ISWC 2025). https://doi.org/10.24406/publica-5924

Genet, Asefa, , and Chen Zongxiong, , and Oussama Zoubia, , and Fidan Limani, , and Kanishka Silva, , and Muhammad Asif Suryani, , and Benjamin Zapilko, , and Leyla Jael Castro, , and Ekaterina Kutafina, , and Dhwani Solanki, , and Heike Fliegl, , and Sonja Schimmler, , and Zeyd Boukhers, , and Harald Sack. "A survey on metadata for Machine Learning models and datasets: standards, practices, and harmonization challenges." Sci-K Workshop at the International Semantic Web Conference (ISWC 2025), 2025.

Genet, Asefa, , and Chen Zongxiong, , and Oussama Zoubia, , and Fidan Limani, , and Kanishka Silva, , and Muhammad Asif Suryani, , and Benjamin Zapilko, , and Leyla Jael Castro, , and Ekaterina Kutafina, , and Dhwani Solanki, , and Heike Fliegl, , and Sonja Schimmler, , and Zeyd Boukhers, , and Harald Sack. "A survey on metadata for Machine Learning models and datasets: standards, practices, and harmonization challenges." Sci-K Workshop at the International Semantic Web Conference (ISWC 2025) (2025).
@inproceedings{Genet2025A, title = {A survey on metadata for Machine Learning models and datasets: standards, practices, and harmonization challenges}, author = {Asefa Genet and Chen Zongxiong and Oussama Zoubia and Fidan Limani and Kanishka Silva and Muhammad Asif Suryani and Benjamin Zapilko and Leyla Jael Castro and Ekaterina Kutafina and Dhwani Solanki and Heike Fliegl and Sonja Schimmler and Zeyd Boukhers and Harald Sack}, booktitle = {Sci-K Workshop at the International Semantic Web Conference (ISWC 2025)}, year = {2025}, doi = {10.24406/publica-5924}, url = {https://ceur-ws.org/Vol-4065/paper05.pdf}, publisher = {ISWC Workshop Proceedings}, }
The growing availability of machine learning (ML) models, datasets, and related artifacts across platforms, such as Hugging Face, GitHub, and Zenodo, has amplified the need for structured and standardized metadata. However, metadata practices remain highly heterogeneous, differing in schema design, vocabulary usage, and semantic expressiveness, posing significant challenges for tasks such as representation, extraction, alignment, and integration. This fragmentation impedes the development of infrastructures that depend on machine-actionable metadata to support discovery, provenance tracking, or cross-platform interoperability. While metadata is also foundational to enabling FAIR principles in ML, there is a lack of consolidated understanding of how existing standards support interoperability and alignment across platforms. In this survey, we review and compare a range of general-purpose and ML-specific metadata standards, evaluating their suitability for cross-platform alignment, discoverability, extensibility, and interoperability.
Multi-Objective Reinforcement Learning for Large Language Model Optimization: Visionary Perspective

reinforcement learning large language models optimization multi-objective RL

Lingxiao Kong , Cong Yang , Oya Beyan , Zeyd Boukhers

Multi-Objective Decision Making Workshop at ECAI 2025 2025

PDF DOI
Kong, L., Yang, C., Beyan, O., & Boukhers, Z. (2025, September). Multi-Objective Reinforcement Learning for Large Language Model Optimization: Visionary Perspective. Multi-Objective Decision Making Workshop at ECAI 2025. https://doi.org/arXiv.2509.21613

Kong, Lingxiao, , and Cong Yang, , and Oya Beyan, , and Zeyd Boukhers. "Multi-Objective Reinforcement Learning for Large Language Model Optimization: Visionary Perspective." Multi-Objective Decision Making Workshop at ECAI 2025, 2025.

Kong, Lingxiao, , and Cong Yang, , and Oya Beyan, , and Zeyd Boukhers. "Multi-Objective Reinforcement Learning for Large Language Model Optimization: Visionary Perspective." Multi-Objective Decision Making Workshop at ECAI 2025 (2025).
@inproceedings{Kong2025Multi-Objective, title = {Multi-Objective Reinforcement Learning for Large Language Model Optimization: Visionary Perspective}, author = {Lingxiao Kong and Cong Yang and Oya Beyan and Zeyd Boukhers}, booktitle = {Multi-Objective Decision Making Workshop at ECAI 2025}, year = {2025}, doi = {arXiv.2509.21613}, url = {https://arxiv.org/pdf/2509.21613}, publisher = {ECAI Workshop Proceedings}, }
Multi-Objective Reinforcement Learning (MORL) presents significant challenges and opportunities for optimizing multiple objectives in Large Language Models (LLMs). We introduce a MORL taxonomy and examine the advantages and limitations of various MORL methods when applied to LLM optimization, identifying the need for efficient and flexible approaches that accommodate personalization functionality and inherent complexities in LLMs and RL. We propose a vision for a MORL benchmarking framework that addresses the effects of different methods on diverse objective relationships. As future research directions, we focus on meta-policy MORL development that can improve efficiency and flexibility through its bi-level learning paradigm, highlighting key research questions and potential solutions for improving LLM performance.
FAIR Data Publisher: Strategie für den flächendeckenden Einsatz von FAIR Digital Objects

FAIR data FAIR Digital Objects data infrastructure data interoperability data reuse data accessibility research data management European data ecosystem

Marius Bensley , Philippe Calvez , Christian Backe , Patrick Draheim , Veit Briken , Anna Petschulies , Fabian Rack , Franziska Boehm , Jan Nagler , Christina Gillmann , Christoph Lange-Bever , Daham Mustafa , Joshua Gelhaas , Maik Mannsfeld , Michael Steinert , Zeyd Boukhers , Jana Böhm , Philipp Wieder , Sonja Sternkopf , Sven Bingert , Suat Can , Henrik Wörden , Insa Nüske , Sina Rohde , Timm Fitschen , Oliver Rack , Frank Tristram , Ashley Caselli , Philipp Essen , Tobias Kuhn , Virginia Balseiro , Ziroli Plutschow , Igor Garmaev , Marius Politze , Torben Miny , David Wagner , Hans-Günther Döbereiner , Lennard Scheurer , Lukas Schmidt , Malte Ohmstеdes , Robert Porzel , Torben Klarl , Mario Martini , Tom Schuchort , Nils Niendorf

Mission KI Report 2025

PDF DOI
Bensley, M., Calvez, P., Backe, C., Draheim, P., Briken, V., Petschulies, A., Rack, F., Boehm, F., Nagler, J., Gillmann, C., Lange-Bever, C., Mustafa, D., Gelhaas, J., Mannsfeld, M., Steinert, M., Boukhers, Z., Böhm, J., Wieder, P., Sternkopf, S., … Niendorf, N. (2025). FAIR Data Publisher: Strategie für den flächendeckenden Einsatz von FAIR Digital Objects. Mission KI Report. https://doi.org/10.24406/publica-6827

Bensley, Marius, , and Philippe Calvez, , and Christian Backe, , and Patrick Draheim, , and Veit Briken, , and Anna Petschulies, , and Fabian Rack, , and Franziska Boehm, , and Jan Nagler, , and Christina Gillmann, , and Christoph Lange-Bever, , and Daham Mustafa, , and Joshua Gelhaas, , and Maik Mannsfeld, , and Michael Steinert, , and Zeyd Boukhers, , and Jana Böhm, , and Philipp Wieder, , and Sonja Sternkopf, , and Sven Bingert, , and Suat Can, , and Henrik Wörden, , and Insa Nüske, , and Sina Rohde, , and Timm Fitschen, , and Oliver Rack, , and Frank Tristram, , and Ashley Caselli, , and Philipp Essen, , and Tobias Kuhn, , and Virginia Balseiro, , and Ziroli Plutschow, , and Igor Garmaev, , and Marius Politze, , and Torben Miny, , and David Wagner, , and Hans-Günther Döbereiner, , and Lennard Scheurer, , and Lukas Schmidt, , and Malte Ohmstеdes, , and Robert Porzel, , and Torben Klarl, , and Mario Martini, , and Tom Schuchort, , and Nils Niendorf. "FAIR Data Publisher: Strategie für den flächendeckenden Einsatz von FAIR Digital Objects." Mission KI Report, 2025.

Bensley, Marius, , and Philippe Calvez, , and Christian Backe, , and Patrick Draheim, , and Veit Briken, , and Anna Petschulies, , and Fabian Rack, , and Franziska Boehm, , and Jan Nagler, , and Christina Gillmann, , and Christoph Lange-Bever, , and Daham Mustafa, , and Joshua Gelhaas, , and Maik Mannsfeld, , and Michael Steinert, , and Zeyd Boukhers, , and Jana Böhm, , and Philipp Wieder, , and Sonja Sternkopf, , and Sven Bingert, , and Suat Can, , and Henrik Wörden, , and Insa Nüske, , and Sina Rohde, , and Timm Fitschen, , and Oliver Rack, , and Frank Tristram, , and Ashley Caselli, , and Philipp Essen, , and Tobias Kuhn, , and Virginia Balseiro, , and Ziroli Plutschow, , and Igor Garmaev, , and Marius Politze, , and Torben Miny, , and David Wagner, , and Hans-Günther Döbereiner, , and Lennard Scheurer, , and Lukas Schmidt, , and Malte Ohmstеdes, , and Robert Porzel, , and Torben Klarl, , and Mario Martini, , and Tom Schuchort, , and Nils Niendorf. "FAIR Data Publisher: Strategie für den flächendeckenden Einsatz von FAIR Digital Objects." Mission KI Report (2025).
@article{Bensley2025FAIRDataPublisher, title = {FAIR Data Publisher: Strategie für den flächendeckenden Einsatz von FAIR Digital Objects}, author = {Marius Bensley and Philippe Calvez and Christian Backe and Patrick Draheim and Veit Briken and Anna Petschulies and Fabian Rack and Franziska Boehm and Jan Nagler and Christina Gillmann and Christoph Lange-Bever and Daham Mustafa and Joshua Gelhaas and Maik Mannsfeld and Michael Steinert and Zeyd Boukhers and Jana Böhm and Philipp Wieder and Sonja Sternkopf and Sven Bingert and Suat Can and Henrik Wörden and Insa Nüske and Sina Rohde and Timm Fitschen and Oliver Rack and Frank Tristram and Ashley Caselli and Philipp Essen and Tobias Kuhn and Virginia Balseiro and Ziroli Plutschow and Igor Garmaev and Marius Politze and Torben Miny and David Wagner and Hans-Günther Döbereiner and Lennard Scheurer and Lukas Schmidt and Malte Ohmstеdes and Robert Porzel and Torben Klarl and Mario Martini and Tom Schuchort and Nils Niendorf}, journal = {Mission KI Report}, year = {2025}, doi = {10.24406/publica-6827}, url = {https://mission-ki.de/media/pages/files/e6eb8545eb-1763376569/fair-data-publisher_bericht_ds.pdf}, }
Der FAIR Data Publisher stellt eine zentrale technologische Initiative dar, um die Auffindbarkeit, Zugänglichkeit, Interoperabilität und Nachnutzbarkeit von Daten bedeutend zu verbessern. Dieses Projekt ist nicht nur ein technisches Werkzeug, sondern trägt zu einer wichtigen Entwicklung in der deutschen und europäischen Dateninfrastruktur bei, die von einem Konsortium aus mehreren namhaften Organisationen gemeinsam vorangetrieben wird.

2024

Falcon 7b for software mention detection in scholarly documents

software mention detection scholarly publications natural language processing large language models class imbalance adaptive sampling

Ameerali Khan , Qusai Ramadan , Cong Yang , Zeyd Boukhers

ESWC Workshop on Natural Scientific Language Processing and Research Knowledge Graphs 2024

PDF DOI
Khan, A., Ramadan, Q., Yang, C., & Boukhers, Z. (2024, June). Falcon 7b for software mention detection in scholarly documents. ESWC Workshop on Natural Scientific Language Processing and Research Knowledge Graphs. https://doi.org/10.1007/978-3-031-65794-8_20

Khan, Ameerali, , and Qusai Ramadan, , and Cong Yang, , and Zeyd Boukhers. "Falcon 7b for software mention detection in scholarly documents." ESWC Workshop on Natural Scientific Language Processing and Research Knowledge Graphs, 2024.

Khan, Ameerali, , and Qusai Ramadan, , and Cong Yang, , and Zeyd Boukhers. "Falcon 7b for software mention detection in scholarly documents." ESWC Workshop on Natural Scientific Language Processing and Research Knowledge Graphs (2024).
@inproceedings{Khan2024Falcon, title = {Falcon 7b for software mention detection in scholarly documents}, author = {Ameerali Khan and Qusai Ramadan and Cong Yang and Zeyd Boukhers}, booktitle = {ESWC Workshop on Natural Scientific Language Processing and Research Knowledge Graphs}, year = {2024}, doi = {10.1007/978-3-031-65794-8_20}, url = {https://doi.org/10.1007/978-3-031-65794-8_20}, publisher = {Springer}, }
This paper investigates applying Falcon-7b (a large language model) for detecting and classifying software mentions in scholarly texts. Focusing on Subtask I of Software Mention Detection in Scholarly Publications (SOMD), the authors explore training strategies including a dual-classifier approach, adaptive sampling, and weighted loss scaling to improve detection accuracy despite class imbalance and the complex syntax of academic writing. Their experiments demonstrate that selective labelling and adaptive sampling yield improved performance – though combining all strategies does not necessarily aggregate gains – and highlight challenges in using LLMs for scholarly text analysis.

2023

Pedestrian collision prediction using a monocular camera

pedestrian detection collision prediction monocular camera ADAS 3D detection computer vision

Shiyuan Chen , Xue Qin , Zeyd Boukhers , John See , Wei Sui , Cong Yang

8th International Workshop on Sensor-Based Activity Recognition and Artificial Intelligence (iWOAR 2023) 2023

PDF DOI
Chen, S., Qin, X., Boukhers, Z., See, J., Sui, W., & Yang, C. (2023, October). Pedestrian collision prediction using a monocular camera. 8th International Workshop on Sensor-Based Activity Recognition and Artificial Intelligence (IWOAR 2023). https://doi.org/10.1145/3615834.3615846

Chen, Shiyuan, , and Xue Qin, , and Zeyd Boukhers, , and John See, , and Wei Sui, , and Cong Yang. "Pedestrian collision prediction using a monocular camera." 8th International Workshop on Sensor-Based Activity Recognition and Artificial Intelligence (iWOAR 2023), 2023.

Chen, Shiyuan, , and Xue Qin, , and Zeyd Boukhers, , and John See, , and Wei Sui, , and Cong Yang. "Pedestrian collision prediction using a monocular camera." 8th International Workshop on Sensor-Based Activity Recognition and Artificial Intelligence (iWOAR 2023) (2023).
@inproceedings{Chen2023Pedestrian, title = {Pedestrian collision prediction using a monocular camera}, author = {Shiyuan Chen and Xue Qin and Zeyd Boukhers and John See and Wei Sui and Cong Yang}, booktitle = {8th International Workshop on Sensor-Based Activity Recognition and Artificial Intelligence (iWOAR 2023)}, year = {2023}, doi = {10.1145/3615834.3615846}, url = {https://doi.org/10.1145/3615834.3615846}, publisher = {ACM}, }
This paper introduces a simple yet efficient method, PedView, for pedestrian collision warning in Advanced Driver Assistance Systems (ADAS). In contrast to existing approaches that rely on LiDAR and stereo cameras for pedestrian-vehicle distance calculation, our proposed PedView leverages a forward-looking monocular camera for 3D pedestrian detection – suitable for resource-limited environments like dealer-installed ADAS. PedView not only estimates distance and speed but also derives pedestrian location and intent via a 3D detector and fuzzy rules, enabling end-to-end collision prediction. Experiments on two datasets demonstrate significant improvements vs. discrete conditional rules (Precision 0.937 vs. 0.844, Recall 0.835 vs. 0.746), highlighting robustness across real-world scenarios.

2022

Evaluation scheme of FAIRness in scholarly data

FAIR data scholarly metadata open citations data evaluation data quality FAIRness metrics

Zeyd Boukhers

Workshop on Open Citations and Open Scholarly Metadata 2022

PDF DOI
Boukhers, Z. (2022, July). Evaluation scheme of FAIRness in scholarly data. Workshop on Open Citations and Open Scholarly Metadata. https://doi.org/10.5281/zenodo.7133290

Boukhers, Zeyd. "Evaluation scheme of FAIRness in scholarly data." Workshop on Open Citations and Open Scholarly Metadata, 2022.

Boukhers, Zeyd. "Evaluation scheme of FAIRness in scholarly data." Workshop on Open Citations and Open Scholarly Metadata (2022).
@inproceedings{Boukhers2022Evaluation, title = {Evaluation scheme of FAIRness in scholarly data}, author = {Zeyd Boukhers}, booktitle = {Workshop on Open Citations and Open Scholarly Metadata}, year = {2022}, doi = {10.5281/zenodo.7133290}, url = {https://doi.org/10.5281/zenodo.7133290}, publisher = {Zenodo / Workshop Proceedings}, }
This paper proposes an evaluation scheme for measuring FAIRness (Findability, Accessibility, Interoperability, Reusability) in scholarly data. The scheme provides a structured methodology for assessing the extent to which scholarly metadata and citation data comply with FAIR principles. The paper discusses the metrics and evaluation criteria used, and demonstrates the scheme on a set of scholarly data repositories.
NFDI4DataScience Registry for Reproducible Data Science and Artificial Intelligence

reproducible research FAIR data data infrastructure metadata knowledge graphs data science AI

Dietrich Rebholz-Schuhmann , Leyla Jael Castro , Zeyd Boukhers , Olga Giraldo , Adamantios Koumpis , Oya Beyan

Workshop on Open Citations and Open Scholarly Metadata 2022

PDF DOI
Rebholz-Schuhmann, D., Castro, L. J., Boukhers, Z., Giraldo, O., Koumpis, A., & Beyan, O. (2022, July). NFDI4DataScience Registry for Reproducible Data Science and Artificial Intelligence. Workshop on Open Citations and Open Scholarly Metadata. https://doi.org/10.5281/zenodo.7129715

Rebholz-Schuhmann, Dietrich, , and Leyla Jael Castro, , and Zeyd Boukhers, , and Olga Giraldo, , and Adamantios Koumpis, , and Oya Beyan. "NFDI4DataScience Registry for Reproducible Data Science and Artificial Intelligence." Workshop on Open Citations and Open Scholarly Metadata, 2022.

Rebholz-Schuhmann, Dietrich, , and Leyla Jael Castro, , and Zeyd Boukhers, , and Olga Giraldo, , and Adamantios Koumpis, , and Oya Beyan. "NFDI4DataScience Registry for Reproducible Data Science and Artificial Intelligence." Workshop on Open Citations and Open Scholarly Metadata (2022).
@inproceedings{Rebholz-Schuhmann2022NFDI4DataScience, title = {NFDI4DataScience Registry for Reproducible Data Science and Artificial Intelligence}, author = {Dietrich Rebholz-Schuhmann and Leyla Jael Castro and Zeyd Boukhers and Olga Giraldo and Adamantios Koumpis and Oya Beyan}, booktitle = {Workshop on Open Citations and Open Scholarly Metadata}, year = {2022}, doi = {10.5281/zenodo.7129715}, url = {https://doi.org/10.5281/zenodo.7129715}, publisher = {Zenodo / Workshop Proceedings}, }
The paper describes the vision and objectives of the NFDI4DataScience (NFDI4DS) consortium: to support the entire research data lifecycle for data science and AI in Germany. It proposes infrastructure for storage and sharing of data, code, models, metadata, and knowledge graphs to support reproducibility, transparency, and compliance with FAIR principles.

2021

LaHAR: Latent Human Activity Recognition using LDA

human activity recognition HAR unsupervised learning LDA sensory data sensor-based HAR topic modeling

Zeyd Boukhers , Danniene Wete , Steffen Staab

The AAAI 2021 Workshop on Plan, Activity, and Intent Recognition 2021

PDF
Boukhers, Z., Wete, D., & Staab, S. (2021, February). LaHAR: Latent Human Activity Recognition using LDA. The AAAI 2021 Workshop on Plan, Activity, and Intent Recognition. https://arxiv.org/abs/2011.11151

Boukhers, Zeyd, , and Danniene Wete, , and Steffen Staab. "LaHAR: Latent Human Activity Recognition using LDA." The AAAI 2021 Workshop on Plan, Activity, and Intent Recognition, 2021.

Boukhers, Zeyd, , and Danniene Wete, , and Steffen Staab. "LaHAR: Latent Human Activity Recognition using LDA." The AAAI 2021 Workshop on Plan, Activity, and Intent Recognition (2021).
@inproceedings{Boukhers2021LaHAR, title = {LaHAR: Latent Human Activity Recognition using LDA}, author = {Zeyd Boukhers and Danniene Wete and Steffen Staab}, booktitle = {The AAAI 2021 Workshop on Plan, Activity, and Intent Recognition}, year = {2021}, url = {https://arxiv.org/abs/2011.11151}, publisher = {AAAI Workshop Proceedings}, }
Processing sequential multi-sensor data becomes important in many tasks due to the dramatic increase in the availability of sensors that can acquire sequential data over time. Human Activity Recognition (HAR) is one such field. Unlike most approaches which use predefined activity classes, this paper proposes a novel approach to discover latent HAR patterns in sequential data. We employ Latent Dirichlet Allocation (LDA) – originally a topic-modeling method – applied to "sensory words" extracted from the sequential sensor data, turning sensor sequences into Bag-of-Words representations. Experiments on a challenging HAR dataset show that LDA can uncover underlying structures in the data, producing human-understandable representations; extrinsic evaluations demonstrated accurate clustering of sequences comparable to labelled activities.
Early detection of COVID lies without evidences and propagation information

COVID-19 misinformation detection fake news NLP early detection social media health misinformation

Ipek Baris , Zeyd Boukhers

CONSTRAINT @ AAAI2021 1-8 2021

PDF
Baris, I., & Boukhers, Z. (2021). Early detection of COVID lies without evidences and propagation information. CONSTRAINT @ AAAI2021, 1–8.

Baris, Ipek, , and Zeyd Boukhers. "Early detection of COVID lies without evidences and propagation information." CONSTRAINT @ AAAI2021, 2021, pp. 1-8.

Baris, Ipek, , and Zeyd Boukhers. "Early detection of COVID lies without evidences and propagation information." CONSTRAINT @ AAAI2021 (2021): 1-8.
@inproceedings{proc16, title = {Early detection of COVID lies without evidences and propagation information}, author = {Ipek Baris and Zeyd Boukhers}, booktitle = {CONSTRAINT @ AAAI2021}, pages = {1-8}, year = {2021}, url = {}, publisher = {ACM}, }
The spread of misinformation and COVID-19 related lies on social media poses a significant public health risk. This paper proposes an approach for early detection of COVID-19 lies and misinformation that does not rely on external evidence or propagation network information, making it applicable at the time of publication before fact-checking resources are available. The approach leverages linguistic and semantic features extracted from the claim text to identify likely misinformation about COVID-19.
Bib2Auth: Deep Learning Approach for Author Disambiguation using Bibliographic Data

author disambiguation deep learning bibliographic data DBLP co-authorship digital libraries

Zeyd Boukhers , Nagaraj Bahubali , Abinaya Thulsi Chandrasekaran , Adarsh Anand , Soniya Manchenahalli Gnanendra Prasad , Sriram Aralappa

BibliODAP workshop @ SIGKDD2021 1-9 2021

PDF
Boukhers, Z., Bahubali, N., Chandrasekaran, A. T., Anand, A., Prasad, S. M. G., & Aralappa, S. (2021). Bib2Auth: Deep Learning Approach for Author Disambiguation using Bibliographic Data. BibliODAP Workshop @ SIGKDD2021, 1–9.

Boukhers, Zeyd, , and Nagaraj Bahubali, , and Abinaya Thulsi Chandrasekaran, , and Adarsh Anand, , and Soniya Manchenahalli Gnanendra Prasad, , and Sriram Aralappa. "Bib2Auth: Deep Learning Approach for Author Disambiguation using Bibliographic Data." BibliODAP workshop @ SIGKDD2021, 2021, pp. 1-9.

Boukhers, Zeyd, , and Nagaraj Bahubali, , and Abinaya Thulsi Chandrasekaran, , and Adarsh Anand, , and Soniya Manchenahalli Gnanendra Prasad, , and Sriram Aralappa. "Bib2Auth: Deep Learning Approach for Author Disambiguation using Bibliographic Data." BibliODAP workshop @ SIGKDD2021 (2021): 1-9.
@inproceedings{proc17, title = {Bib2Auth: Deep Learning Approach for Author Disambiguation using Bibliographic Data}, author = {Zeyd Boukhers and Nagaraj Bahubali and Abinaya Thulsi Chandrasekaran and Adarsh Anand and Soniya Manchenahalli Gnanendra Prasad and Sriram Aralappa}, booktitle = {BibliODAP workshop @ SIGKDD2021}, pages = {1-9}, year = {2021}, url = {}, }
Author disambiguation is a critical challenge in digital libraries where multiple authors share the same name. This paper presents Bib2Auth, a deep learning approach for author disambiguation using bibliographic data. Bib2Auth uses co-authorship networks and publication metadata to create author representations that can be used to link publication records to their correct authors. Experiments conducted in the BibliODAP workshop demonstrate the effectiveness of the approach on DBLP data.

2016

Convoy detection in crowded surveillance videos

convoy detection surveillance video crowd analysis behavior understanding video analytics

Zeyd Boukhers , Yicong Wang , Kimiaki Shirahama , Kuniaki Uehara , Marcin Grzegorzek

Human Behavior Understanding (HBU) at ACM Multimedia International Workshop 2016 2016

PDF DOI
Boukhers, Z., Wang, Y., Shirahama, K., Uehara, K., & Grzegorzek, M. (2016, September). Convoy detection in crowded surveillance videos. Human Behavior Understanding (HBU) at ACM Multimedia International Workshop 2016. https://doi.org/10.1007/978-3-319-46843-3_9

Boukhers, Zeyd, , and Yicong Wang, , and Kimiaki Shirahama, , and Kuniaki Uehara, , and Marcin Grzegorzek. "Convoy detection in crowded surveillance videos." Human Behavior Understanding (HBU) at ACM Multimedia International Workshop 2016, 2016.

Boukhers, Zeyd, , and Yicong Wang, , and Kimiaki Shirahama, , and Kuniaki Uehara, , and Marcin Grzegorzek. "Convoy detection in crowded surveillance videos." Human Behavior Understanding (HBU) at ACM Multimedia International Workshop 2016 (2016).
@inproceedings{Boukhers2016Convoy, title = {Convoy detection in crowded surveillance videos}, author = {Zeyd Boukhers and Yicong Wang and Kimiaki Shirahama and Kuniaki Uehara and Marcin Grzegorzek}, booktitle = {Human Behavior Understanding (HBU) at ACM Multimedia International Workshop 2016}, year = {2016}, doi = {10.1007/978-3-319-46843-3_9}, url = {https://doi.org/10.1007/978-3-319-46843-3_9}, publisher = {Springer International Publishing}, }
This paper proposes detection of convoys in a crowded surveillance video. A convoy is defined as a group of pedestrians who are moving or standing together for a certain period of time. To detect such convoys, we firstly address pedestrian detection in a crowded scene, where small regions of pedestrians and their strong occlusions render usual object detection methods ineffective. Thus, we develop a method that detects pedestrian regions by clustering feature points based on their spatial characteristics. Then, positional transitions of pedestrian regions are analysed by our convoy detection method that consists of the clustering and intersection processes. The former finds groups of pedestrians in one frame by flexibly handling their relative spatial positions, and the latter refines groups into convoys by considering their temporal consistences over multiple frames. The experimental results on a challenging dataset shows the effectiveness of our convoy detection method.

2015

Object detection and depth estimation for 3D trajectory extraction

object detection depth estimation 3D trajectory video analysis multimedia indexing

Zeyd Boukhers , Kimiaki Shirahama , Frédéric Li , Marcin Grzegorzek

International Workshop on Content-Based Multimedia Indexing (CBMI 2015) 2015

PDF DOI
Boukhers, Z., Shirahama, K., Li, F., & Grzegorzek, M. (2015, July). Object detection and depth estimation for 3D trajectory extraction. International Workshop on Content-Based Multimedia Indexing (CBMI 2015). https://doi.org/10.1109/cbmi.2015.7153632

Boukhers, Zeyd, , and Kimiaki Shirahama, , and Frédéric Li, , and Marcin Grzegorzek. "Object detection and depth estimation for 3D trajectory extraction." International Workshop on Content-Based Multimedia Indexing (CBMI 2015), 2015.

Boukhers, Zeyd, , and Kimiaki Shirahama, , and Frédéric Li, , and Marcin Grzegorzek. "Object detection and depth estimation for 3D trajectory extraction." International Workshop on Content-Based Multimedia Indexing (CBMI 2015) (2015).
@inproceedings{Boukhers2015Object, title = {Object detection and depth estimation for 3D trajectory extraction}, author = {Zeyd Boukhers and Kimiaki Shirahama and Frédéric Li and Marcin Grzegorzek}, booktitle = {International Workshop on Content-Based Multimedia Indexing (CBMI 2015)}, year = {2015}, doi = {10.1109/cbmi.2015.7153632}, url = {https://doi.org/10.1109/cbmi.2015.7153632}, publisher = {IEEE}, }
To detect an event which is defined by the interaction of objects in a video, it is necessary to capture their spatio-temporal relation. However, the video only displays the original 3D space which is projected onto a 2D image plane. This paper introduces a method which extracts 3D trajectories of objects from 2D videos. Each trajectory represents the transition of an object’s positions in the 3D space. We extract such trajectories by combining object detection with depth estimation that estimates the depth information in 2D videos. The major problem for this is the inconsistency between object detection and depth estimation results. For example, significantly different depths may be estimated for the region of the same object, and an object region that is appropriately shaped by estimated depths may be missed. To overcome this, we first initialise the 3D position of an object by selecting the frame with the highest consistency between the object detection and depth estimation results. Then, we track the object in the 3D space using particle filter, where the 3D position of this object is modelled as a hidden state to generate its 2D visual appearance. Experimental results demonstrate the effectiveness of our method.

Books & Book Chapters

2017

3D Trajectory Extraction from 2D Videos for Human Activity Analysis

3D trajectory extraction monocular video human activity analysis particle filter camera odometry depth estimation computer vision video analysis

Zeyd Boukhers

44 2017

PDF
Boukhers, Z. (2017). 3D Trajectory Extraction from 2D Videos for Human Activity Analysis (Vol. 44). Logos Verlag Berlin GmbH. https://www.logos-verlag.de/cgi-bin/engbuchmid?isbn=4501&lng=deu&id=

Boukhers, Zeyd. "3D Trajectory Extraction from 2D Videos for Human Activity Analysis." , vol. 44, 2017.

Boukhers, Zeyd. "3D Trajectory Extraction from 2D Videos for Human Activity Analysis." 44 (2017).
@book{book1, title = {3D Trajectory Extraction from 2D Videos for Human Activity Analysis}, author = {Zeyd Boukhers}, volume = {44}, year = {2017}, url = {https://www.logos-verlag.de/cgi-bin/engbuchmid?isbn=4501&lng=deu&id=}, publisher = {Logos Verlag Berlin GmbH}, }
This thesis presents methods for extracting 3D trajectories of objects from 2D videos recorded by monocular cameras, with application to human activity analysis. The approach addresses the challenge of recovering 3D spatial information from 2D video projections without relying on stereo cameras or depth sensors. The book covers particle filter-based trajectory extraction, example-based 3D trajectory learning, camera odometry estimation from monocular video, and applications to human activity recognition in 3D space. Extensive experiments demonstrate the effectiveness of the proposed methods on standard benchmarks.

Preprints & Technical Reports

2025

InstaDa: Augmenting Instance Segmentation Data with Dual-Agent System

instance segmentation data augmentation dual-agent system generative AI synthetic data deep learning

Xianbao Hou , Yonghao He , Zeyd Boukhers , John See , Hu Su , Wei Sui , Cong Yang

arXiv preprint arXiv:2509.02973 2025

PDF DOI
Hou, X., He, Y., Boukhers, Z., See, J., Su, H., Sui, W., & Yang, C. (2025). InstaDa: Augmenting Instance Segmentation Data with Dual-Agent System. ArXiv Preprint ArXiv:2509.02973. https://doi.org/10.48550/arXiv.2509.02973

Hou, Xianbao, , and Yonghao He, , and Zeyd Boukhers, , and John See, , and Hu Su, , and Wei Sui, , and Cong Yang. "InstaDa: Augmenting Instance Segmentation Data with Dual-Agent System." arXiv preprint arXiv:2509.02973, 2025.

Hou, Xianbao, , and Yonghao He, , and Zeyd Boukhers, , and John See, , and Hu Su, , and Wei Sui, , and Cong Yang. "InstaDa: Augmenting Instance Segmentation Data with Dual-Agent System." arXiv preprint arXiv:2509.02973 (2025).
@article{hou2025instada, title = {InstaDa: Augmenting Instance Segmentation Data with Dual-Agent System}, author = {Xianbao Hou and Yonghao He and Zeyd Boukhers and John See and Hu Su and Wei Sui and Cong Yang}, journal = {arXiv preprint arXiv:2509.02973}, year = {2025}, doi = {10.48550/arXiv.2509.02973}, url = {https://arxiv.org/pdf/2509.02973}, }
Instance segmentation requires large amounts of annotated training data, which is expensive and time-consuming to collect. This paper presents InstaDa, a system for augmenting instance segmentation training data using a dual-agent framework. InstaDa employs two complementary agents: a generation agent that creates new synthetic instances using generative models, and a selection agent that evaluates and filters the generated instances for quality and diversity. Experiments demonstrate that InstaDa significantly improves instance segmentation performance when training data is limited.
MoiréNet: A Compact Dual-Domain Network for Image Demoiréing

moiré pattern image demoiréing dual-domain network image restoration deep learning frequency domain

Shuwei Guo , Simin Luan , Yan Ke , Zeyd Boukhers , John See , Cong Yang

arXiv preprint arXiv:2509.18910 2025

PDF DOI
Guo, S., Luan, S., Ke, Y., Boukhers, Z., See, J., & Yang, C. (2025). MoiréNet: A Compact Dual-Domain Network for Image Demoiréing. ArXiv Preprint ArXiv:2509.18910. https://doi.org/10.48550/arXiv.2509.18910

Guo, Shuwei, , and Simin Luan, , and Yan Ke, , and Zeyd Boukhers, , and John See, , and Cong Yang. "MoiréNet: A Compact Dual-Domain Network for Image Demoiréing." arXiv preprint arXiv:2509.18910, 2025.

Guo, Shuwei, , and Simin Luan, , and Yan Ke, , and Zeyd Boukhers, , and John See, , and Cong Yang. "MoiréNet: A Compact Dual-Domain Network for Image Demoiréing." arXiv preprint arXiv:2509.18910 (2025).
@article{guo2025moir, title = {MoiréNet: A Compact Dual-Domain Network for Image Demoiréing}, author = {Shuwei Guo and Simin Luan and Yan Ke and Zeyd Boukhers and John See and Cong Yang}, journal = {arXiv preprint arXiv:2509.18910}, year = {2025}, doi = {10.48550/arXiv.2509.18910}, url = {https://arxiv.org/pdf/2509.18910}, }
Moiré artifacts are a common problem in digital photography when capturing screens or periodic structures. Existing demoiréing methods are often computationally expensive or fail to handle complex moiré patterns. This paper presents MoiréNet, a compact dual-domain network for image demoiréing. MoiréNet processes images in both spatial and frequency domains simultaneously, leveraging complementary information to remove moiré artifacts while preserving image details. Experiments on standard benchmarks demonstrate that MoiréNet achieves competitive performance with significantly fewer parameters than existing methods.
Reinforcement Learning for Large Language Model Fine-Tuning: A Systematic Literature Review

reinforcement learning large language models fine-tuning RLHF DPO PPO LLM alignment systematic review

Lingxiao Kong , Qusai Ramdan , Oussama Zoubia , Jahid Hasan Polash , Mayra Elwes , Mehdi Akbari Gurabi , Lu Jin , Ekaterina Kutafina , Roman Matzutt , Yuanbin Wang , others

2025

PDF DOI
Kong, L., Ramdan, Q., Zoubia, O., Polash, J. H., Elwes, M., Gurabi, M. A., Jin, L., Kutafina, E., Matzutt, R., Wang, Y., & others. (2025). Reinforcement Learning for Large Language Model Fine-Tuning: A Systematic Literature Review. https://doi.org/10.21203/rs.3.rs-8196796/v1

Kong, Lingxiao, , and Qusai Ramdan, , and Oussama Zoubia, , and Jahid Hasan Polash, , and Mayra Elwes, , and Mehdi Akbari Gurabi, , and Lu Jin, , and Ekaterina Kutafina, , and Roman Matzutt, , and Yuanbin Wang, , and others. "Reinforcement Learning for Large Language Model Fine-Tuning: A Systematic Literature Review." , 2025.

Kong, Lingxiao, , and Qusai Ramdan, , and Oussama Zoubia, , and Jahid Hasan Polash, , and Mayra Elwes, , and Mehdi Akbari Gurabi, , and Lu Jin, , and Ekaterina Kutafina, , and Roman Matzutt, , and Yuanbin Wang, , and others. "Reinforcement Learning for Large Language Model Fine-Tuning: A Systematic Literature Review." (2025).
@article{kong2025reinforcement, title = {Reinforcement Learning for Large Language Model Fine-Tuning: A Systematic Literature Review}, author = {Lingxiao Kong and Qusai Ramdan and Oussama Zoubia and Jahid Hasan Polash and Mayra Elwes and Mehdi Akbari Gurabi and Lu Jin and Ekaterina Kutafina and Roman Matzutt and Yuanbin Wang and others}, year = {2025}, doi = {10.21203/rs.3.rs-8196796/v1}, url = {https://assets-eu.researchsquare.com/files/rs-8196796/v1_covered_96aef1fc-9b91-4ac6-8e6d-0c088657015f.pdf}, }
Reinforcement learning (RL) has emerged as a key technique for aligning large language models (LLMs) with human preferences and improving their performance on downstream tasks. This paper presents a systematic literature review of RL methods applied to LLM fine-tuning, covering approaches such as RLHF (Reinforcement Learning from Human Feedback), PPO-based methods, DPO (Direct Preference Optimization), and multi-objective RL. The review analyzes the strengths and limitations of existing approaches, identifies open research challenges, and proposes a taxonomy of RL fine-tuning methods for LLMs.

2024

SuperEdge: Towards a Generalization Model for Self-Supervised Edge Detection

edge detection self-supervised learning generalization computer vision multi-homography pixel-level detection object-level detection

Kai Leng , Zhijie Zhang , Jie Liu , Zeyd Boukhers , Wei Sui , Cong Yang , Zhijun Li

arXiv preprint arXiv:2401.02313 2024

PDF DOI
Leng, K., Zhang, Z., Liu, J., Boukhers, Z., Sui, W., Yang, C., & Li, Z. (2024). SuperEdge: Towards a Generalization Model for Self-Supervised Edge Detection. ArXiv Preprint ArXiv:2401.02313. https://doi.org/10.48550/arXiv.2401.02313

Leng, Kai, , and Zhijie Zhang, , and Jie Liu, , and Zeyd Boukhers, , and Wei Sui, , and Cong Yang, , and Zhijun Li. "SuperEdge: Towards a Generalization Model for Self-Supervised Edge Detection." arXiv preprint arXiv:2401.02313, 2024.

Leng, Kai, , and Zhijie Zhang, , and Jie Liu, , and Zeyd Boukhers, , and Wei Sui, , and Cong Yang, , and Zhijun Li. "SuperEdge: Towards a Generalization Model for Self-Supervised Edge Detection." arXiv preprint arXiv:2401.02313 (2024).
@article{leng2024superedge, title = {SuperEdge: Towards a Generalization Model for Self-Supervised Edge Detection}, author = {Kai Leng and Zhijie Zhang and Jie Liu and Zeyd Boukhers and Wei Sui and Cong Yang and Zhijun Li}, journal = {arXiv preprint arXiv:2401.02313}, year = {2024}, doi = {10.48550/arXiv.2401.02313}, url = {https://arxiv.org/abs/2401.02313}, }
Edge detection is a fundamental task in computer vision, but existing supervised methods require expensive manual annotations. This paper presents SuperEdge, a streamlined yet efficient model for self-supervised edge detection capable of concurrently extracting edges at pixel-level and object-level granularity. SuperEdge eliminates the dependency on manual annotated edge labels through self-supervised training, thereby enhancing its generalizability across diverse datasets. The approach employs a multi-level, multi-homography technique to transfer annotations from synthetic to real-world datasets, achieving competitive performance compared to supervised methods.

2021

Hybrid physics and deep learning model for interpretable vehicle state prediction

vehicle state prediction hybrid model physics-informed deep learning interpretability recurrent neural network motion model

Alexandra Baier , Zeyd Boukhers , Steffen Staab

arXiv preprint arXiv:2103.06727 2021

PDF DOI
Baier, A., Boukhers, Z., & Staab, S. (2021). Hybrid physics and deep learning model for interpretable vehicle state prediction. ArXiv Preprint ArXiv:2103.06727. https://doi.org/10.48550/arXiv.2103.06727

Baier, Alexandra, , and Zeyd Boukhers, , and Steffen Staab. "Hybrid physics and deep learning model for interpretable vehicle state prediction." arXiv preprint arXiv:2103.06727, 2021.

Baier, Alexandra, , and Zeyd Boukhers, , and Steffen Staab. "Hybrid physics and deep learning model for interpretable vehicle state prediction." arXiv preprint arXiv:2103.06727 (2021).
@article{baier2021hybrid, title = {Hybrid physics and deep learning model for interpretable vehicle state prediction}, author = {Alexandra Baier and Zeyd Boukhers and Steffen Staab}, journal = {arXiv preprint arXiv:2103.06727}, year = {2021}, doi = {10.48550/arXiv.2103.06727}, url = {https://arxiv.org/pdf/2103.06727}, }
Physical motion models offer interpretability but require expensive measurements of parameters like aerodynamic and hydrodynamic properties, while recurrent neural networks achieve high accuracy at low cost but their results are hard to interpret. To solve this, the authors propose a hybrid approach combining deep learning and physical motion models including a novel two-phase training procedure. Additionally, they achieve interpretability by restricting the output range of the deep neural network as part of the hybrid model, which limits the uncertainty introduced by the neural network to a known quantity. Experiments on vehicle dynamics datasets demonstrate that the hybrid model achieves comparable accuracy to pure neural network models while providing interpretable predictions.

Publications

Journal Articles

2025

Context-agnostic data visitation to exchange health data for scientific research: the example of PADME PHT

R4-SLAM: Toward Real-time, Robust, and Resource-Restricted Visual SLAM in Dynamic Environments

2024

Skeleton ground truth extraction: methodology, annotation tool and benchmarks

BladeView: Towards Automatic Wind Turbine Inspection with Unmanned Aerial Vehicle

A non-enhanced CT based deep learning diagnostic system for high-risk COVID-19 infection in lung cancer patients

2023

Knowledge guided multi-filter residual convolutional neural network for ICD coding from clinical text

Deep author name disambiguation using DBLP data

2022

COIN: Counterfactual Image Generation for VQA Interpretation

2020

Analyzing the Influence of Hyper-parameters and Regularizers of Topic Modeling in Terms of Renyi Entropy

2019

Fractal approach for determining the optimal number of topics in the field of topic modeling

2017

Example-based 3D trajectories of objects from 2D videos

Less Restrictive Camera Odometry from Monocular Camera

Conference Papers

2026

Unleashing Semantic and Geometric Priors for 3D Scene Completion

Enhancing RL Generalizability in Robotics through SHAP Analysis of Algorithms and Hyperparameters

2025

From Theory to Practice: Demonstrators of FAIR Data Spaces Across Different Sectors

ELMTEX: Fine-tuning large language models for structured clinical information extraction. A case study on clinical reports

Future-Proofing Heritage with ARGUS: A Multimodal Digital Twin Approach for Sustainable Preservation

Q-Chain: A Causal-Aware Framework for Structural and Educational Question Generation

A unified framework for cultural heritage data historicity and migration: The ARGUS approach

EMORL: Ensemble multi-objective reinforcement learning for efficient and flexible LLM fine-tuning

Exploring the limits of model compression in LLMs: a knowledge distillation study on QA tasks

FAIR Data Assessment Using LLMs: The Fair-Way

2024

YawnNet: A Visual-Centric Approach for Yawning Detection

Moiré pattern detection: Stability and efficiency with evaluated loss function

Transition in focus of prediction tasks for skeleton graph component detection with Transformer

Large language model in medical informatics: direct classification and enhanced text representations for automatic ICD coding

FDO manager: minimum viable FAIR digital object implementation

2023

Data trading and monetization: challenges and open research directions

NFDI4DS Infrastructure and Services

Research knowledge graphs in NFDI4DS

FAIR Data Spaces - Building a Common Cloud-Based Data Space for Industry and Science by Linking Gaia-X and NFDI

Beyond trading data: the hidden influence of public awareness and interest on cryptocurrency volatility

Enhancing Reproducibility in Research through FAIR Digital Objects

PADME-SoSci: A Platform for Analytics and Distributed Machine Learning for the Social Sciences

Enhancing data space semantic interoperability through Machine Learning: a visionary perspective

2022

Whois? Deep author name disambiguation using bibliographic data

Vision and Natural Language for Metadata Extraction from PDF Documents: A Multimodal Approach

2021

MexPub: Deep transfer learning for metadata extraction from German publications

BibliODAP’21: The 1st workshop on bibliographic data analysis and processing

2019

An end-to-end approach for extracting and segmenting high-variance references from PDF documents

EXCITE – A toolchain to extract, match and publish open literature references

2016

Automatic recognition of movement patterns in the Vojta therapy using RGB-D data

2015

Shape-Based Eye Blinking Detection and Analysis

Environmental microbiological content-based image retrieval system using internal structure histogram

Extracting 3D Trajectories of Objects from 2D Videos Using Particle Filter

Workshop Papers

2025

MESD: Metadata Extraction from Scholarly Documents - A Shared Task Overview

Towards Systematic Specification and Verification of Fairness Requirements: A Position Paper

A survey on metadata for Machine Learning models and datasets: standards, practices, and harmonization challenges

Multi-Objective Reinforcement Learning for Large Language Model Optimization: Visionary Perspective

FAIR Data Publisher: Strategie für den flächendeckenden Einsatz von FAIR Digital Objects

2024

Falcon 7b for software mention detection in scholarly documents

2023

Pedestrian collision prediction using a monocular camera

2022

Evaluation scheme of FAIRness in scholarly data

NFDI4DataScience Registry for Reproducible Data Science and Artificial Intelligence

2021

LaHAR: Latent Human Activity Recognition using LDA