SentimentalITsts at SemEval-2016 Task 4: building a Twitter sentiment analyzer in your backyard

Proceedings of SemEval-2016, pages 243–246, San Diego, California, June 16-17, 2016. c⃝2016 Association for Computational Linguistics SentimentalITsts at SemEval-2016 Task 4: building a Twitter sentiment analyzer in your backyard Cosmin Florean, Oana Bejenaru, Eduard Apostol, Octavian Ciobanu, Adrian Iftene and Diana Trandabăţ University “Alexandru Ioan Cuza” of Iaşi, Romania {cosmin.florean, oana.bejenaru, eduard.apostol, octavian.ciobanu, adiftene, dtrandabat}@info.uaic.ro Abstract The paper presents the system developed by the SentimentalITsts team for the participation in Semeval-2016 task 4, in the subtasks A, B and C. The developed system uses off the shelf solutions for the development of a quick sentiment analyzer for tweets. However, the lack of any syntactic or semantic information resulted in performances lower than those of other teams. 1 Introduction Slowly but surely, social media replaced the tradi- tional sources of information: people’s need to be constantly updated changed our behavior from buying a newspaper or watching TV, to using a Facebook or Twitter account to visualize, in a cus- tomizable manner, the day’s hottest news, with the bonus of being able to also comment on them. Social media sites gained their popularity due to the “freedom” of expression they induce in peo- ple’s mind: being able to post real time messages about your opinions on whatever topic you come across, discuss political and social decisions, com- plain, express gratitude or exchange impressions about products you use in everyday life. Texts shared through social media applications offer us the information that we need: for example, the reviews of a product provide us useful infor- mation about its advantages and disadvantages, while the text of an advertisement invites us to eat at the new Chinese restaurant in town. As huge amounts of texts become available through social media, a challenging task concerns the organization and processing of this information to extract knowledge. Natural language processing tools trained on large news corpora have usually problems when applied to unstandardized social media inputs, mainly due to the fact that social media content can appear in various forms (Becker et al., 2012), from photos and video updates to news, offers and literary works, and various infor- mal formats. Twitter is micro-blogging platform where peo- ple can send messages to one or multiple users, fol- low friends and read messages without much diffi- culty. Twitter messages, commonly known as tweets, are limited to 140 characters, and frequent- ly include hashtags (labels which should make it easier for users to find messages with similar con- tent), all in one making Twitter analysis charming. Out of the 5 subtasks of Semeval-2016 task 4, the SentimentalITsts participated in subtask A: Message Polarity Classification , subtask B Tweet classification according to a two-point scale and subtask C Tweet classification according to a five- point scale. Subtask A asked to classify a given tweet in three categories: positive, negative or neu- tral, according to the identified sentiment (Nakov et al., 2016). The tweet was known to be about a specific topic (by topic is meant anything people usually express opinions about on …

Content truncated to 500 words