Dr. Sebastian Arnold is a visiting researcher at DATEXIS at Beuth University of Applied Sciences Berlin. He graduated from Université de Fribourg with a PHD in 2020 and with a Master  in computer science at Technische Universität Berlin in 2014 and has worked on search intentions, entity linking, topic classification and text segmentation. He is one of the main developers of the information extraction framework TeXoothe factual search engine GoOLAP and the interactive TASTY (Tag-as-you-type) editor. Off-the-record, he enjoys manifold activities as musician, drummer and hardware hacker. Sebastian is currently writing his thesis about neural machine reading for domain-specific text resources.

Research Interests

  • Neural document representations for machine reading
  • Local topic and named entity extraction
  • Answer passage retrieval
  • Deep learning / machine learning

Projects

  • CDV Contextual Discourse Vectors
  • SECTOR Neural model for Coherent Topic Segmentation and Classification
  • WikiSection Dataset for cliniclal topics in long documents
  • TeXoo Java framework for text analytics with Deep Learning
  • TASTY Interactive Entity Linking "Tag-as-you-type"
  • GoOLAP Factual search engine
  • Nerdle Topic-expert Question Answering system
  • Senode Interactive music sequencer app

Publications (see DBLP)

  • Jens-Michalis Papaioannou, Manuel Mayrdorfer, Sebastian Arnold, Felix A. Gers, Klemens Budde and Alexander Löser: Aspect-based Passage Retrieval with Contextualized Discourse Vectors, Proceedings of the 43rd European Conference on Information Retrieval ECIR 2021
  • Sebastian Arnold, Betty van Aken, Paul Grundmann, Felix A. Gers and Alexander Löser. Learning Contextualized Document Representations for Healthcare Answer Retrieval.The Web Conference 2020 (WWW'20). ACM, 2020: 1332–1343 [PDF] [code].
  • Sebastian Arnold, Rudolf Schneider, Philippe Cudré-Mauroux, Felix A. Gers and Alexander Löser. SECTOR: A Neural Model for Coherent Topic Segmentation and Classification.Transactions of the Association for Computational Linguistics (TACL) Vol. 7. MIT Press, 2019: 169-184. [PDF] [code] [dataset] [slides]
  • Rudolf Schneider, Sebastian Arnold, Tom Oberhauser, Tobias Klatt, Thomas Steffek and Alexander Löser: Smart-MD: Neural Paragraph Retrieval of Medical Topics.World Wide Web Conference (Companion). IW3C2, 2018: 203–206 [PDF]
  • Sebastian Arnold, Robert Dziuba and Alexander Löser: TASTY: Interactive Entity Linking As-You-Type.26th International Conference on Computational Linguistics (COLING'16) Demos. ACL, 2016: 111–115 [PDF] [demo]
  • Sebastian Arnold, Felix A. Gers, Torsten Kilias and Alexander Löser: Robust Named Entity Recognition in Idiosyncratic Domains. arXiv:1608.06757 [cs.CL] 2016 [PDF] [code]
  • Sebastian Arnold, Alexander Löser and Torsten Kilias: Resolving Common Analytical Tasks in Text Databases. ACM Eighteenth International Workshop On Data Warehousing and OLAP. ACM 2015: 75–84 [PDF]
  • Umar Maqsud, Sebastian Arnold, Michael Hülfenhaus and Alan Akbik: Nerdle: Topic-Specific Question Answering Using Wikia Seeds.25th International Conference on Computational Linguistics: Demos. ACM 2014: 81–85 [PDF] [demo]
  • Sebastian Arnold, Damian Burke, Tobias Dörsch, Bernd Löber and Andreas Lommatzsch: News Visualization based on Semantic Knowledge.International Semantic Web Conference (Posters & Demos) 2014: 5–8 [PDF]
  • Alexander Löser, Sebastian Arnold and Tillmann Fiehn: The GoOLAP Fact Retrieval Framework.Lecture Notes in Business Information Processing Vol 96, Business Intelligence. Springer Berlin Heidelberg, 2012: 84–97 [PDF] [demo]

Contact

E-Mail: sarnold (at) bht-berlin.de

Twitter: @sebastianarnold

https://www.xing.com/profile/Sebastian_Arnold18

https://www.linkedin.com/in/sebastianarnold/