SINAI: Voting System for Twitter Sentiment Analysis

Proceedings of the 8th International Workshop on Semantic Evaluation (SemEval 2014), pages 572–577, Dublin, Ireland, August 23-24, 2014. SINAI: Voting System for Twitter Sentiment Analysis Eugenio Mart´ınez-C´amara, Salud Mar´ıa Jim´enez-Zafra, M. Teresa Mart´ın-Valdivia, L. Alfonso Ure˜na-L´opez SINAI Research Group University of Ja´en E-23071, Ja´en (Spain) {emcamara, sjzafra, maite, laurena}@ujaen.es Abstract This article presents the participation of the SINAI research group in the task Sen- timent Analysis in Twitter of the SemEval Workshop. Our proposal consists of a voting system of three polarity classifiers which follow a lexicon-based approach. 1 Introduction Opinion Mining (OM) or Sentiment Analysis (SA) is the task focuses on the computational treatment of opinion, sentiment and subjectivity in texts (Pang and Lee, 2008). Currently, OM is a trendy task in the field of Natural Language Processing due mainly to the fact of the growing interest in the knowledge of the opinion of people from dif- ferent sectors of the society. The interest in the research community for the extraction of the sentiment in Twitter posts is re- flected in the organization of several workshops with the aim of promoting the research in this task. Two are the most relevant, the first is the task Sentiment Analysis in Twitter celebrated within the SemEval workshop whose first edition was in 2013 (Nakov et al., 2013). The second is the work- shop TASS1 , which is a workshop for promot- ing the research in sentiment analysis in Spanish in Twitter. The first edition of the workshop took place in 2012 (Villena-Rom´an et al., 2013). The 2014 edition of the task Sentiment Analy- sis in Twitter proposes a first subtask, which has as challenge the sentiment classification at entity level, and a second subtask that consists of the polarity classification at document or tweet level. The training corpus is the same than the former This work is licensed under a Creative Commons At- tribution 4.0 International Licence. Page numbers and pro- ceedings footer are added by the organisers. Licence details: http://creativecommons.org/licenses/by/4.0/ 1http://www.daedalus.es/TASS edition, but this year the test corpus is consider- ably bigger than the prior one. A wider description of the task and the corpus can be read in (Rosen- thal et al., 2014). We present an unsupervised polarity classifica- tion system for the subtask B of the task Senti- ment Analysis in Twitter. The system is based on a voting strategy of three lexicon-based senti- ment classifiers. The sentiment analysis research community broadly knows the lexicons selected. They are, SentiWordNet (Baccianella et al., 2010), the lexicon developed by Hu and Liu (Hu and Liu, 2004) and the MPQA lexicon (Wilson et al., 2005). The rest of the paper is organized as follows. The following section focuses on the description of the different sentiment resources used for de- veloping the sentiment classifiers. The subsequent section outlines the system proposed for the 2014 edition of the task. The last section exposes the analysis of the results reached this year. 2 Sentiment lexical resources Sentiment lexicons are lexical resources com- posed of …

Content truncated to 500 words