Proceedings of the 2008 Conference on Semantics in Text Processing - STEP '08

Abstract

Thanks to both statistical approaches and finite state methods, natural language processing (NLP), particularly in the area of robust, open-domain text processing, has made considerable progress in the last couple of decades. It is probably fair to say that NLP tools have reached satisfactory performance at the level of syntactic processing, be the output structures chunks, phrase structures, or dependency graphs. Therefore, the time seems ripe to extend the state-of-the-art and consider deep semantic processing as a serious task in wide-coverage NLP. This is a step that normally requires syntactic parsing, as well as integrating named entity recognition, anaphora resolution, thematic role labelling and word sense disambiguation, and other lower levels of processing for which reasonably good methods have already been developed. The goal of the STEP workshop is to provide a forum for anyone active in semantic processing of text to discuss innovative technologies, representation issues, inference techniques, prototype implementations, and real applications. The preferred processing targets are large quantities of texts---either specialised domains, or open domains such as newswire text, blogs, and wikipedia-like text. Implemented rather than theoretical work is emphasised in STEP. Featuring in STEP 2008 workshop is a "shared task" on comparing semantic representations as output by state-of-the-art NLP systems. Participants were asked to supply a (small) text, before the workshop. The test data for the shared task is composed out of all the texts submitted by the participants, allowing participants to "challenge" each other. The output of these systems will be judged on a number of aspects by a panel of experts in the field, during the workshop.