Sebastian Varges
Background     Research Interests     Projects     Ph.D. thesis    Publications 

Dr. Sebastian Varges
Department of Information and Communication Technology
University of Trento
Room D11, via Sommarive 14
38050 Povo di Trento, Italy

phone:    +39  0461 88 2081
email: varges  +  AT + disi.unitn.it


Background:  I am a Marie Curie Senior Fellow at the Adaptive Multimodal Information and  Interfaces  Research Lab at the University of Trento (Italy), where I work on the ADAMACH project.  Until December 2006, I was a Research Engineer at the Computational Semantics Lab at the CSLI, Stanford University.   Before that, I was at the ITRI, University of Brighton where I worked on the TUNA project.  I conducted my Ph.D. research at the ICCS, University of Edinburgh on "Instance-based Natural Language Generation".  I obtained an undergraduate degree in Computational Linguistics from Saarbrücken University and a Master's degree from the University of Düsseldorf.

I am a board member of the Special Interest Group in Natural Language Generation (SIGGEN) from 2009-2013.

Research Interests (there are many, but here is a selection):
  • Dialogue systems, natural language generation (NLG), and natural language processing in general:  NLG may turn out to be the key for natural language understanding, especially in the context of dialogue systems, if we want to understand what the other has said - and why.
  • Combining rule-based and machine learning techniques such as instance-based learning and reinforcement learning:  can we combine the strengths of both but avoid their disadvantages? I am particularly interested in working with small amounts of training data and rule systems of limited complexity.
  • Lexicalized grammar formalisms, and their application to dialogue (beyond the syntactic level).
  • The role of content and task knowledge in dialogue processing.
  • NLP system development, programming systems, and annotation environments: we are still developing most research software from scratch - what languages and programming environments are most suitable for collaborative software development?
Current and past projects:
  • Adaptive Multimodal Information and  Interfaces  Research Lab (ADAMACH project) at the University of Trento (Italy):  Jan 2006 - current:  we are developing new models of dialogue and speech processing, that, for example, learn during interaction. 
  • Computational Semantics Lab at the CSLI, Stanford University (June 2005 - Dec  2006):  I worked on the generation component of an In-car Multi-agent Dialogue System that is based on the CSLI dialogue manager.  The system operates in three domains of increasing difficulty:  1) playing songs from an MP3 database, 2) discussing what restaurants are available in the area, 3) generating route descriptions.
  • TUNA (Oct. 2003 - May 2005):  In the TUNA project at the former ITRI, Brighton, we developed algorithms for referring expression generation (GRE). I was particularly interested in applying the "overgeneration and ranking" approach I had previously used for surface realization in NLG to a higher-level content determination task such as GRE.
  • BEETLE (University of Edinburgh, Jan. - Sept. 2003):  I developed an XSLT-based generator for the BEETLE tutorial dialogue system.
  • M-Piro (University of Edinburgh):  I worked on enhancing the expressive capabilities of M-Piro, the Java version of the ILEX generator, for a few months in 2002.
  • Computational Linguistics project at the University of Düsseldorf, headed by James Kilbury: between 1994 and 1998 I
Ph.D. thesis:
Publications:
  • Sebastian Varges, Silvia Quarteroni, Giuseppe Riccardi, Alexei V. Ivanov, Pierluigi Roberti:   Leveraging POMDPs trained with User Simulations and Rule-based Dialogue Management in a Spoken Dialogue System.   Demo paper to appear in Proceedings of The 10th Annual SIGDIAL Meeting on Discourse and Dialogue (SIGDIAL), London, UK, 11-12 September, 2009.
  • 2006. Fuliang Weng, Sebastian Varges, Badri Raghunathan, Florin Ratiu, Heather Pon-Barry, Brian Lathrop, Qi Zhang, Tobias Scheideck, Harry Bratt, Kui Xu, Matthew Purver, Rohit Mishra, Madhuri Raya, Stanley Peters, Yao Meng, Lawrence Cavedon and Liz Shriberg: CHAT: A Conversational Helper for Automotive Tasks.  Proceedings of the 9th International Conference on Spoken Language Processing (Interspeech), Pittsburgh, PA, September 2006.
  • 2004. Johanna Moore, Kaska Porayska-Pomsta, Sebastian Varges, and Claus Zinn: Generating Tutorial Feedback with Affect. Proceedings of the Seventeenth International Florida Artificial Intelligence Research Symposium Conference (FLAIRS), AAAI Press, 2004.
  • 2001.  Sebastian Varges and Chris Mellish: Instance-based Natural Language Generation. Proceedings of the 2nd Meeting of the North American Chapter of the Association for Computational Linguistics (NAACL-01), pages 1-8. Carnegie Mellon University, Pittsburgh, PA, USA. [also available as postscript]