Collecting and Evaluating Lexical Polarity with A Game With a Purpose

Proceedings of Recent Advances in Natural Language Processing, pages 329–337, Hissar, Bulgaria, Sep 7–9 2015. Collecting and Evaluating Lexical Polarity with a Game With A Purpose Mathieu Lafourcade LIRMM Univ. Montpellier, France lafourca@lirmm.fr Nathalie Le Brun Imagin@t 34400 Lunel, France imaginat@imaginat.name Alain Joubert LIRMM Univ. Montpellier, France aj@lirmm.fr Abstract Sentiment analysis from a text requires amongst others having a polarity lexical resource. We designed LikeIt, a GWAP (Game With A Purpose) that allows to at- tribute a positive, negative or neutral value to a term, and thus obtain a resulting polar- ity for most of the terms of the freely avail- able lexical network of the JeuxDeMots project. We present a quantitative analy- sis of data obtained through our approach, together with the comparison method we developed to validate them qualitatively. 1 Introduction Being able to evaluate feelings is essential in nat- ural language processing, whether to analyze po- litical speeches or opinion of the general pub- lic on the provision of services, tourist, cultural, or about consumer goods. Whatever type of ap- proach, statistics supervised or more linguistic one (Brun, 2011), such ability requires referring to a polarity lexical resource, in wich terms are en- dowed with positive, negative and neutral values. The polarity can be expressed using a single nu- merical value (Taboada et al., 2011), or more: two values (positive/negative) are used in Emolex (Saif and Turney, 2013), a lexical polarity / feelings resource in English, produced by crowdsourcing (using Amazon Mechanical Turk, which can be problematic, see Fort et al. (2014)). SentiWord- Net (Esuli and Sebastiani, 2006), as well as Word- Net Affect (Strapparava and Valitutti, 2004) are extensions of WorldNet in which terms are polar- ized along three values (positive, negative, objec- tive) of which the last one is opposed to the first two. Approaches through propagation by starting from a manual core (see Gala and Brun (2012) and Lafourcade and Fort (2014)) have also been per- formed, but such approaches may not exactly re- flect the views of speakers. Learning algorithms and compositional approaches can use such polar- ity data (Kim and Hovy, 2004) and (Turney, 2002). The GWAP (Game With A Purpose) JeuxDeMots (JDM) (Lafourcade, 2007) resulted in a lexical network in constant expansion, in which terms are linked by lexical-semantic relations. Contributive approaches with non-experts have been analysed in (Snow et al., 2013) and proven quite efficient. Within the JDM project, some alternative type of games allow to validate and verify lexical relations produced through the main game (Lafourcade et al., 2015). The JDM project thus provides suitable context to test various methods of polarity infor- mation acquisition. It may be relevant to assign to words some in- formation in the form of finite sets of values. Thus, the polarity can be defined by three values: posi- tive, negative and neutral. It may be noticed that many semantic features can be characterized in this way, i.e. associated with such variable-sized sets of values: feelings/emotions (anger, fear, joy, love, sadness ...), or colors …

Content truncated to 500 words