How do computers determine the meaning of individual words in a text, and what challenges do they face with deliberately ambiguous usages such as puns?
Machine learning approaches to word sense disambiguation (WSD) depend on the availability of large numbers of training examples, which can be expensive or impractical to obtain. This is a particular problem for processing the sort of lexical-semantic anomalies employed for deliberate effect in humour and wordplay. In contrast to machine learning systems are knowledge-based techniques, which rely only on pre-existing lexical-semantic resources (LSRs) such as dictionaries and thesauri. In this talk, we treat the task of improving the performance and applicability of knowledge-based WSD, both generally and for the particular case of wordplay. In the first part of the talk, we present two approaches for bridging the "lexical gap" problem and thereby improving WSD coverage and accuracy. In the first approach, we supplement the word's context and the LSR's sense descriptions with entries from a distributional thesaurus. The second approach enriches an LSR's sense information by aligning and clustering the senses to those of other, complementary LSRs. In the second part of the talk, we describe how these techniques, along with evaluation methodologies from traditional WSD, can be adapted for the "disambiguation" of puns, or rather for the automatic identification of their double meanings. We conclude with a sketch of how this and other techniques from computational semantics could be used to help translate puns from one language to another.
Time: Thursday, 13th of June 2019, 6:30 p.m. sharp
Location: Oesterreichisches Forschungsinstitut fuer Artificial Intelligence (OFAI), Freyung 6, Stiege 6, 1010 Wien