Twitter Based System: Using Twitter for Disambiguating Sentiment Ambiguous Adjectives

Proceedings of the 5th International Workshop on Semantic Evaluation, ACL 2010, pages 436–439, Uppsala, Sweden, 15-16 July 2010. c⃝2010 Association for Computational Linguistics Twitter Based System: Using Twitter for Disambiguating Sentiment Ambiguous Adjectives Alexander Pak, Patrick Paroubek Universit´e de Paris-Sud, Laboratoire LIMSI-CNRS, Bˆatiment 508, F-91405 Orsay Cedex, France alexpak@limsi.fr, pap@limsi.fr Abstract In this paper, we describe our system which participated in the SemEval 2010 task of disambiguating sentiment ambigu- ous adjectives for Chinese. Our system uses text messages from Twitter, a popu- lar microblogging platform, for building a dataset of emotional texts. Using the built dataset, the system classifies the meaning of adjectives into positive or negative sen- timent polarity according to the given con- text. Our approach is fully automatic. It does not require any additional hand-built language resources and it is language in- dependent. 1 Introduction The dataset of the SemEval task (Wu and Jin, 2010) consists of short texts in Chinese contain- ing target adjectives whose sentiments need to be disambiguated in the given contexts. Those adjec- tives are: 大 big, 小 small, 多 many, 少 few, 高 high, 低 low, 厚 thick, 薄 thin, 深 deep, shallow, 重 heavy, light, 巨大 huge, 重大 grave. Disambiguating sentiment ambiguous adjec- tives is a challenging task for NLP. Previous stud- ies were mostly focused on word sense disam- biguation rather than sentiment disambiguation. Although both problems look similar, the latter is more challenging in our opinion because impreg- nated with more subjectivity. In order to solve the task, one has to deal not only with the semantics of the context, but also with the psychological as- pects of human perception of emotions from the written text. In our approach, we use Twitter1 microblogging platform to retrieve emotional messages and form two sets of texts: messages with positive emotions and those with negative ones (Pak and Paroubek, 1http://twitter.com 2010). We use emoticons2 as indicators of an emo- tion (Read, 2005) to automatically classify texts into positive or negative sets. The reason we use Twitter is because it allows us to collect the data with minimal supervision efforts. It provides an API3 which makes the data retrieval process much more easier then Web based search or other re- sources. After the dataset of emotional texts has been obtained, we build a classifier based on n-grams Na¨ıve Bayes approach. We tested two approaches to build a sentiment classifier: 1. In the first one, we collected Chinese texts from Twitter and used them to train a classi- fier to annotate the test dataset. 2. In the second one, we used machine trans- lator to translate the dataset from Chinese to English and annotated it using collected En- glish texts from Twitter as the training data. We have made the second approach because we were able to collect much more of English texts from Twitter than Chinese ones and we wanted to test the impact of machine translation on the performance of our classifier. We have exper- imented with Google Translate and Yahoo Ba- …

Content truncated to 500 words