22 November 2013, Computational Social Choice Seminar, Umberto Grandi

Abstract

Sentiment analysis assigns a positive, negative or neutral polarity to an item or entity, extracting individual opinions from text by means of natural language processing tools. Current sentiment analysis techniques are satisfactory in case there is a single entity under consideration, but can lead to inaccurate or wrong results when dealing with a set of possibly correlated items. In this talk I will argue in favor of using techniques from voting theory and preference aggregation to provide accurate definitions of the collective sentiment for a set of multiple items. In particular, I will analyse possible data structures that can be extracted from a corpus of individual expressions, and propose an innovative procedure to define the collective sentiment. This procedure is built on the classical Borda count, and uses both the polarity and the preference information that can be extracted from a corpus of textual individual expressions. Joint work with Andrea Loreggia and Francesca Rossi.

For more information, see http://www.illc.uva.nl/~ulle/seminar/ or contact Ulle Endriss (ulle.endriss at uva.nl).