Wissenschaftlicher Mitarbeiter / Research Assistant
Department of Linguistics
Abstract anaphora refers to a variety of anaphora in which certain expressions (e.g. this, that, it, etc.) are used to refer to abstract objects, such as facts, propositions, or events. Such abstract objects are generally represented syntactically by verbal phrases or whole sentences, which renders them difficult to handle automatically, since many of the features commonly used for anaphora resolution are not available. This project has as its goal a detailed quantitative investigation of the properties of abstract anaphora as well as the construction of computational systems capable of processing such expressions adequately. I also plan to investigate to what degree resolving abstract anaphora can improve performance in other NLP tasks, such as automatic summarization or translation.
Kolhatkar, V., Roussel, A., Dipper, S. and Zinsmeister, H. (2018). Anaphora with Non-nominal Antecedents in Computational Linguistics: A Survey. Computational Linguistics, 44(3). pp. 547–612. [pdf]
Roussel, A., Dipper, S., Jablotschkin, S. and Zinsmeister, H. (2018). Towards the Automatic Resolution of Discourse Deixis: Insights from Annotation. Proceedings of the 14th Conference on Natural Language Processing (KONVENS). pp. 178–191. Vienna, Austria. [pdf]
Roussel, A. (2018). Detecting and Resolving Shell Nouns in German. Proceedings of the NAACL-Workshop on Computational Models of Reference, Anaphora, and Coreference (CRAC). pp. 61–67. New Orleans, LA, USA. [pdf]
Simonjetz, F. and Roussel, A. (2016). Crosslinguistic Annotation of German and English Shell Noun Complexes. Proceedings of the 13th Conference of Natural Language Processing (KONVENS). pp. 265–278. Bochum, Germany. [pdf]