:: Home | Print ::

Katja Filippova - Ph.D. candidate

  • Contact Information
EML Research gGmbH
Schloss-Wolfsbrunnenweg 33
69118 Heidelberg
Germany

email: Lastname at eml-r.org
Tel.: +49 (0)6221 - 533 - 238
Fax: +49 (0)6221 - 533 - 298
 
 
 
 
Research Interests
  • generation and summarization
  • discourse coherence and information structure
  • cognitive and functional linguistics
I am working on automatic biography summarization in German. My interests lie in the field of computational linguistics rather than natural language processing, although I constantly use the benefits of machine learning. Ideally, I would like to build a system with a sound linguistic basis which would generate user-oriented biographies from input text(s).
During the first of year of my studies I have focused on local coherence, an essential property of texts generated by humans. Within this topic, I am intrigued by the interplay of word order and salience, accessibility and pronominalization. The data I am working with is in German, but I would like to see if the phenomena found in German also hold for languages with more/less rigid word order (English vs. Russian).

Publications

2007

  • German Vorfeld and Local Coherence | Special Issue on Coherence in Dialogue and Generation of Journal of Logic, Language, and Information (JoLLI). Volume 16(4), pp. 465-485. (Abstract)
    Katja Filippova and Michael Strube

  • Generating Constituent Order in German Clauses | Proceedings of the 45th Annual Meeting of the Association for Computational Linguistics (ACL 07). Prague, Czech Republic, June 23-25, 2007, pp. 320-327. (PDF)
    Katja Filippova and Michael Strube

  • Extending the Entity-grid Coherence Model to Semantically Related Entities | Proceedings of the 11th European Workshop on Natural Language Generation (ENLG 07). Schloss Dagstuhl, Germany, June 17-20, 2007, pp. 139-142. (PDF)
    Katja Filippova and Michael Strube

  • Cascaded Filtering for Topic-Driven Multi-Document Summarization | Proceedings of the 2007 Document Understanding Conference (DUC 07). Rochester, N.Y., April 22-27, 2007. (PDF)
    Katja Filippova, Margot Mieskes, Vivi Nastase, Simone Paolo Ponzetto and Michael Strube

2006

  • Improving Text Fluency by Reordering of Constituents | Proceedings of the ESSLLI Workshop on Modelling Coherence for Generation and Dialogue Systems. Malaga, Spain, July 31 - August 11, 2006, pp. 9-16. (PDF)
    Katja Filippova and Michael Strube

  • Using Linguistically Motivated Features for Paragraph Segmentation | Proceedings of the 2006 Conference on Empirical Methods in Natural Language Processing (EMNLP 06). Sydney, Australia, July 22-23, 2006, pp. 267-274. (PDF)
    Katja Filippova and Michael Strube

2005

  • What Treebanks Can Do for You: Rule-based and Machine-learning Approaches to Anaphora Resolution in German | Proceedings of the 4th Workshop on Treebanks and Linguistic Theories (TLT 05). Barcelona, Spain, December 9-10, 2005, pp. 77-88. (PDF)
    Erhard Hinrichs, Katja Filippova and Holger Wunsch

  • A Data-driven Approach to Pronominal Anaphora Resolution in German | Proceedings of the 5th International Conference on Recent Advances in Natural Language Processing (RANLP 05). Borovets, Bulgaria, September 21-23, 2005, pp. 239-245. (PDF)
    Erhard Hinrichs, Katja Filippova and Holger Wunsch

About Me

Before coming to EML, I studied at school and at University in St.Petersburg (BA in Linguistics), for the MA degree in CL in Tübingen (under supervision of Prof.Dr. Erhard Hinrichs and Dr. Sandra Kübler). Currently, I am a PhD student of Prof.Dr. Elke Teich at TU Darmstadt.

In case you wonder what the best time to go to Petersburg is, I suggest June, late August-early September, and January. If you have to choose between Heidelberg and Tübingen, I would recommend visiting Heidelberg. Even if this information is of no use for you, well, you still have learned something about me.

Personal Interests

I am pretty much interested in everything, in particular, good books, good music, good pictures, and juggling (although I am still clumsy with four). It is tempting to write more, so I might put a list with exact names and titles of what I consider "good".


<< back

© EML Research gGmbH