Sebastian Gombert

Sebastian Gombert started to work as a doctoral researcher at the Educational Technologies group, DIPF, in 2021. Before this, he studied Linguistic and Literary Computing at TU Darmstadt which he finished with a thesis on the detection of semantic uncertainty in scientific writing through a combination of transformer language models and conditional random fields. Besides this, Sebastian worked as a tutor, software developer and junior research assistant in various contexts and was active in academic self-government.

• Natural Language Processing
• Learning Analytics
• Applied Corpus- and Psycholinguistics
• Digital Humanities
• Artificial Intelligence in Education
• Network Science
• Human Computer Interaction

• Master of Arts in Linguistic and Literary Computing, 2021, TU Darmstadt
• Bachelor of Arts in Digital Philology and Computer Science, 2019, TU Darmstadt

• 2014 – Developer @ Sanofi-Aventis Deutschland GmbH
• 2015-2017 – Developer @ Abius GmbH
• 2015-2020 – Co-founder @ mediacollective UG
• 2017 – IT Administrator @ Student Council FB2, TU Darmstadt
• 2016-2021 Tutor and Junior Research Assistant @ Corpus- and Computational Linguistics, TU Darmstadt
• 2020 Tutor and Teaching Assistant @ Ubiquitous Knowledge Processing Lab, TU Darmstadt
• 2020-2021 Developer @ AskAlbert / studiumdigitale, Goethe Universität Frankfurt

Workshop Papers

• Gombert, S. (2021). Twin BERT Contextualized Sentence Embedding Space Learning and Gradient-Boosted Decision Tree Ensembles for Scene Segmentation in German Literature. In A. Zehe, L. Konle, L. K. Dümpelmann, E. Gius, S. S. Guhr, A. Hotho, F. Jannidis, L. Kaufmann, M. Krug, F. Puppe, N. Reiter, & A. Schreiber (Eds.), Proceedings of the Shared Task on Scene Segmentation co-located with the 17th Conference on Natural Language Processing (KONVENS 2021), Düsseldorf, Germany, September 6th, 2021 (Vol. 3001, pp. 42–48). CEUR-WS.org.
• Gombert, S., & Bartsch, S. (2021). TUDA-CCL at SemEval-2021 Task 1: Using Gradient-boosted Regression Tree Ensembles Trained on a Heterogeneous Feature Set for Predicting Lexical Complexity. In Proceedings of the 15th International Workshop on Semantic Evaluation (SemEval-2021) (pp. 130-137). Association for Computational Linguistics.
• Gombert, S. & Bartsch, S. (2020). MultiVitaminBooster at PARSEME Shared Task 2020: Combining Window- and Dependency-Based Features with Multilingual Contextualised Word Embeddings for VMWE Detection. In Proceedings of the Joint Workshop on Multiword Expressions and Electronic Lexicons (pp. 149–155). Association for Computational Linguistics.

Talks/Presentations/Workshops given

• Gombert, S. (2022). Workshop: Assessing constructed responses with explainable natural language processing. In Sixteenth EATEL Summer School on Technology Enhanced Learning.
• Gombert, S., Bhattacharya, S., Di Mitri, D., Drachsler, H. (2022). Interpretierbare maschinelle Sprachverarbeitung für die Identifikation konzeptionellen Wissens in energiephysikbezogenen deutschsprachigen Kurzantworten. In Komplexität nicht nur bewältigen, sondern zu Nutze machen: Log- und Textdaten im Educational Assessment. 9. Tagung der Gesellschaft für Empirische Bildungsforschung.