A Language Independent Method for Generating Large Scale Polarity Lexicons

A Language Independent Method for Generating Large Scale Polarity Lexicons Giuseppe Castellucci, Danilo Croce, Roberto Basili Dept. of Electronic Engineering, Dept. of Enterprise Engineering, Dept. of Enterprise Engineering University of Roma, Tor Vergata, Via del Politecnico 1, 00133 Roma, Italy castellucci@ing.uniroma2.it, croce@info.uniroma2.it, basili@info.uniroma2.it Abstract Sentiment Analysis systems aims at detecting opinions and sentiments that are expressed in texts. Many approaches in literature are based on resources that model the prior polarity of words or multi-word expressions, i.e. a polarity lexicon. Such resources are defined by teams of annotators, i.e. a manual annotation is provided to associate emotional or sentiment facets to the lexicon entries. The development of such lexicons is an expensive and language dependent process, making their coverage of linguistic sentiment phenomena limited. Moreover, once a lexicon is defined it can hardly be adopted in a different language or even a different domain. In this paper, we present several Distributional Polarity Lexicons (DPLs), i.e. large-scale polarity lexicons acquired with an unsupervised methodology based on Distributional Models of Lexical Semantics. Given a set of heuristically annotated sentences from Twitter, we transfer the sentiment information from sentences to words. The approach is mostly unsupervised, and experimental evaluations on Sentiment Analysis tasks in two languages show the benefits of the generated resources. The generated DPLs are publicly available in English and Italian. Keywords: Polarity Lexicon Generation, Sentiment Analysis, Distributional Models 1. Introduction Sentiment Analysis (SA) and Opinion Mining systems (Pang and Lee, 2008) aim at tracking the opinions ex- pressed in texts with respect to specific topics, e.g. prod- ucts or people. In particular, SA deals with the problem of deciding whether a portion of a text, e.g. a sentence or a phrase, is expressing a trend towards specific feelings, e.g. positivity or negativity. Polarity lexicons are list of words each associated to one or more values, indicating their trend towards specific feelings, e.g. a single value in [−1, +1] can be used to indicate negativity (−1), neu- trality (0) or positivity (+1). Such lexicons have been de- fined to support the development of automatic systems for detecting subjective phrases or sentences and recognizing their polarity (Mohammad and Turney, 2010). For exam- ple, in these lexicons, “good” can be associated to a prior positive sentiment in contrast to “sad”, considered negative in every domain. The occurrence of such words in a sen- tence can be adopted as an indicator of the polarity trends that are expressed in the sentence. These lexicons are often hand-compiled; however, from a linguistic point of view, a-priori membership of words to polarity classes can be considered too restrictive, as sentiment expressions are of- ten topic dependent: the occurrences of the word mouse, for example, are mostly neutral in the consumer electron- ics domain, while it can be negatively biased in a restaurant domain. Representing topic specific polarity phenomena would require manual revisions of the lexicon entries, thus resulting in a costly operation. Moreover, these resources mainly exist for the English language, while they scarce for the others. …

Content truncated to 500 words