INESC-ID: A Regression Model for Large Scale Twitter Sentiment Lexicon Induction

INESC-ID: A Regression Model for Large Scale Twitter Sentiment Lexicon Induction Silvio Amir, Ramon F. Astudillo, Wang Ling, Bruno Martins†, M´ario Silva, Isabel Trancoso Instituto de Engenharia de Sistemas e Computadores Investigac¸˜ao e Desenvolvimento Rua Alves Redol 9 Lisbon, Portugal {samir, ramon.astudillo, wlin, mjs, isabel.trancoso}@inesc-id.pt †bruno.g.martins@tecnico.ulisboa.pt Abstract We present the approach followed by INESC- ID in the SemEval 2015 Twitter Sentiment Analysis challenge, subtask E. The goal was to determine the strength of the association of Twitter terms with positive sentiment. Using two labeled lexicons, we trained a regression model to predict the sentiment polarity and in- tensity of words and phrases. Terms were rep- resented as word embeddings induced in an unsupervised fashion from a corpus of tweets. Our system attained the top ranking submis- sion, attesting the general adequacy of the pro- posed approach. 1 Introduction Sentiment lexicons are one of the key resources for the automatic analysis of opinions, emotive and sub- jective text (Liu, 2012). They compile words an- notated with their prior polarity of sentiment, re- gardless of the context. For instance, words such as beautiful or amazing tend to express a positive sen- timent, whereas words like boring or ugly are con- sidered negative. Most sentiment analysis systems use either word count methods, based on sentiment lexicons, or rely on text classifiers. In the former, the polarity of a message is estimated by computing the ratio of (positive and negative) sentiment bear- ing words. Despite its simplicity, this method has been widely used (O’Connor et al., 2010; Bollen and Mao, 2011; Mitchell et al., 2013). Even more so- phisticated systems, based on supervised classifica- tion, can be greatly improved with features derived from lexicons (Kiritchenko et al., 2014). However, manually created sentiment lexicons consist of few carefully selected words. Consequently, they fail to capture the use of non-conventional word spelling and slang, commonly found in social media. This problem motivated the creation of a task in the SemEval 2015 Twitter Sentiment Analysis chal- lenge. This task (subtask E), intended to evaluate au- tomatic methods of generating Twitter specific sen- timent lexicons. Given a set of words or phrases, the goal was to assign a score between 0 and 1, re- flecting the intensity and polarity of sentiment these terms express. Participants were asked to submit a list, with the candidate terms ranked according to sentiment score. This list was then compared to a ranked list obtained from human annotations and the submissions were evaluated using the Kendall (1938) Tau rank correlation metric. In this paper, we describe a system developed for this challenge, based on a novel method to create large scale, domain-specific sentiment lexicons. The task is addressed as a regression problem, in which terms are represented as word embeddings, induced from a corpus of 52 million tweets. Then, using manually annotated lexicons, a regression model was trained to predict the polarity and intensity of sentiment of any word or phrase from that corpus. We found this approach to be effective for the …

Content truncated to 500 words