UMCC_DLSI: Sentiment Analysis in Twitter using Polirity Lexicons and Tweet Similarity

Proceedings of the 8th International Workshop on Semantic Evaluation (SemEval 2014), pages 727–731, Dublin, Ireland, August 23-24, 2014. UMCC DLSI: Sentiment Analysis in Twitter using Polirity Lexicons and Tweet Similarity Pedro Aniel S´anchez-Mirabal, Yarelis Ruano Torres, Suilen Hern´andez Alvarado University of Matanzas / Cuba pedroasm@umcc.cu yara@umcc.cu suilen.alvarado@umcc.cu Yoan Guti´errez, Andr´es Montoyo, Rafael Mu˜noz University of Alicante/Spain ygutierrez@dlsi.ua.es montoyo@dlsi.ua.es rafael@dlsi.ua.es Abstract This paper describes a system sub- mitted to SemEval-2014 Task 4B: Sentiment Analysis in Twitter, by the team UMCC DLSI Sem integrated by researchers of the University of Matanzas, Cuba and the University of Alicante, Spain. The system adopts a cascade classification process that uses two classi- fiers, K-NN using the lexical Levenshtein metric and a Dagging model trained over attributes extracted from annotated cor- pora and sentiment lexicons. Phrases that fit the distance thresholds were automat- ically classified by the KNN model, the others, were evaluated with the Dagging model. This system achieved over 52.4% of correctly classified instances in the Twitter message-level subtask. 1 Introduction Nowadays, one of the most important sources of data to extract useful and heterogeneous knowl- edge is Textual Information. Daily, millions of Tweets, SMS and blog comments increase the huge volume of information available for re- searchers. Texts can provide factual information, such as: descriptions, lists of characteristics, or even instructions to opinion-based information, which would include reviews, emotions, or feel- ings (Guti´errez et al., 2013). These facts have motivated that dealing with the identification and extraction of opinions and sentiments in texts re- quires special attention. Applications of Senti- ment Analysis are now more common than ever in fields like politics and business. More than 50 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/ systems participating in this task, clearly indicate the increase of interest in the scientific community. Twitter messages can be found among of the most used corpora nowadays for Sentiment Anal- ysis (SA). This kind of messages involves an evi- dent informality which has been addressed in dif- ferent ways. For example, there are some works like (Guti´errez et al., 2013) that apply normali- sation textual tools to reduce the informality of the twitter messages. Authors such as (Go et al., 2009), (Guti´errez et al., 2013), (Fern´andez et al., 2013) and others are focused on the application of preprocessing processes and feature reduction to be able to standardise twitter messages and re- duce different types of elements like hashtags, user nicks, urls, etc. In terms of those techniques that can be used for SA, we can cite (Pang et al., 2002) who built a lexicon with associated polarity value, starting with a set of classified seed adjectives and using conjunctions (and) disjunctions (or, but) to deduce the orientation of new words in a corpus. This re- search was based on machine learning techniques to address Sentiment Classification. Other inter- esting research is (Turney, 2002), which classi- fies words according to …

Content truncated to 500 words