Meaning in Context (MIC) Symposium, Munich, September 14 - 18, 2015

Center for Advanced Studies (CAS), LMU University of Munich

Organizers: Hinrich Schütze, Hans Kamp, Alessandro Lenci, James Pustejovsky

Meaning in Context:
Models of Computation and Representation

Building on earlier work in distributional semantics, several new approaches to learning word representations have recently been proposed and shown to perform well on a number of natural language processing tasks. However, one unsolved problem for this approach to semantics is how to model context. There has been some limited success in modeling the contextual effect of topic, the arguably coarsest way of modeling context. But generally, the representation of a word in distributional models is not context-dependent and is combined with other relevant information in an ad hoc manner that varies from task to task.

Following the Dagstuhl Seminar organized in November 2013 on this topic, the present symposium will bring together leading researchers from several disciplines to address the problem of context effects in semantic models of computation and representation. A particular focus will be on formal semantics, since this field has produced a comprehensive body of work modeling such phenomena as: compositionality, interpreting context, inference, and ambiguity.