Frame-based Lexical Semantic Resources for Information Access

Manfred Pinkal 　

Abstract:
The recent progress in information access technology is impressive (see, e.g., the performance of the Google search engine). However, there are obvious limitations to knowledge-lean information access techniques: There is an upper bound of accuracy, which excludes high-quality applications, and these techniques rely on the enormous redundancy of the Web, and thus are unsuited for closed domains with smaller document sizes. One of the most urgent problems to be cared about is the lexical semantics bottleneck, the unavailability of domain-independent lexica with rich semantic information on lexical items. Providing large-scale lexical semantic repositories is an enormous challenge, due to the size of the vocabulary and the inherent complexity and vagueness of lexical meaning.