April 4, 2006
One of the most critical problems in word disambiguation research is determining the sense distinctions that are required or meaningful for natural language processing applications. In recent years, computational linguistics have typically taken a practical approach to the problem, including the use of pre-defined sense inventories in WordNet and other lexical resources, exploration of clustering and similar techniques to identify associated word occurrences that can be regarded as representing a unique sense, and definition of relevant sense distinctions by examining cross-lingual lexicalization of ambiguous words.
Although research on lexical ambiguity has a long history in the field of psycholinguistics, work in this area is rarely considered in debates concerned with identifying relevant sense distinctions for NLP. Recent research in the field of psycholinguistics may offer significant insight into the sense distinctions that humans readily recognize. In addition, psycholinguistics research has much to say about the ways in which humans process ambiguous words and sentences, which can in turn shed some light on computational approaches to lexical disambiguation.
This workshop aims to bring together computational linguists and psycholinguists in order to bring the work in each field to bear on the question of sense distinction. Papers dealing with sense distinction will be solicited from researchers in both fields—in particular, we seek to include papers from psycholinguists that provide insight into the kinds and degree of sense distinction made by humans, and from computational linguists on the problem of identifying appropriate sense distinctions to serve language processing applications.
Christiane Fellbaum, Princeton University (USA)