Marie Bexte

Marie Bexte is a Research Assistant at the FernUniversität in Hagen's Computational Linguistics professorship since April 2022. She previously worked at the Language Technology Lab at the University of Duisburg-Essen (2021–2022). Holding an MSc in Applied Cognitive and Media Sciences with a focus on Cognition and Artificial Intelligence (2021), and a BSc in the same field (2018), her work centers on automated content scoring and image-based task analysis.

• Education:
  • MSc Applied Cognitive and Media Science (Cognition and Artificial Intelligence), 2021
  • BSc Applied Cognitive and Media Science, 2018

Her research focuses on developing and evaluating methods for automated assessment in educational contexts, particularly examining text similarity approaches, zero-shot learning for content scoring, and visio-linguistic model sensitivity to color naming. She co-developed the LeSpell spelling error benchmark and investigates confidence-based grading systems to reduce manual effort.

Key publication trends include:

• Advancing BERT/S-BERT for educational text scoring
• Multimodal analysis of image-text interactions
• Argument mining and prompt sensitivity in writing
• Spelling error detection in learner language
• Cold-start problem solutions using LLM-generated data

She participates in major NLP/educational application workshops (BEA 2022–2025), LREC-COLING, and ACL conferences. Her work bridges computational linguistics with practical classroom applications through systematic benchmarking and model evaluation.