Technical devices are getting more versatile and sophisticated all the time. To make their use more intuitive, the NLP group develops software facilitating the multimodal dialogue between users and machines, applying both statistical and symbolic methods. The group's research focus lies on the semantics and pragmatics of discourse.

ANNOUNCEMENTS

The JAIR paper written by Michael and Simone in 2007 on "Knowledge Derived from Wikipedia for Computing Semantic Relatedness" will receive the 2010 Honorable Mention for the 2010 IJCAI-JAIR Best Paper Prize! Here is the blurb.

Cai and Michael have an exciting paper on a global model for coreference resolution at COLING this year: End-to-End Coreference Resolution via Hypergraph Partitioning.

Viola has a paper on Detecting Vagueness in Two Languages and Two Domains at KOVENS 2010.

Cai and Michael have a paper on coreference resolution evaluation metrics at SIGdial '10: Evaluation Metrics for End-to-End Coreference Resolution Systems. The paper describes evaluation metrics which can deal with system mentions.

Our latest project is CoSyne: A project on Multi-lingual Content Synchronization with Wikis -- an ICT-STREP project funded by the European Commision. The project will start in March 2010. More news later!

We are building WikiNet -- a multi-lingual large scale ontology based on Wikipedia. Currently it contains approximately 3 million concepts and 20+ million relations extracted from the 20090703 version of Wikipedia. We also have a multi-lingual dictionary extracted from the Wiktionary (it contains only entries that have at least one translation).

WikiNet is described in our LREC paper.

Less new news here.