MDSENT at SemEval-2016 Task 4: A Supervised System for Message Polarity Classification

Proceedings of SemEval-2016, pages 139–144, San Diego, California, June 16-17, 2016. c⃝2016 Association for Computational Linguistics MDSENT at SemEval-2016 Task 4: A Supervised System for Message Polarity Classification Hang Gao and Tim Oates hanggao1@umbc.edu, oates@cs.umbc.edu University of Maryland Baltimore County 11000 Hilltop Circlet Baltimore, MD 21250, USA Abstract This paper describes our system submitted for the Sentiment Analysis in Twitter task of SemEval-2016, and specifically for the Mes- sage Polarity Classification subtask. We used a system that combines Convolutional Neural Networks and Logistic Regression for senti- ment prediction, where the former makes use of embedding features while the later utilizes various features like lexicons and dictionaries. 1 Introduction Recently, rapid growth of the amount of user- generated content on the web prompts increasing in- terest in research on sentiment analysis and opinion mining. A typical example is Twitter, where lots of users express feelings and opinions about vari- ous subjects. However, unlike traditional media, lan- guage used in social network services like Twitter is often informal, leading to new challenges to corre- sponding text analysis. The SemEval-2016 Sentiment Analysis in Twit- ter task (SESA-16) is a task that focuses on the sentiment analysis of tweets. As a continuation of SemEval-2015 Task 10, SESA-16 introduces sev- eral new challenges, including the replacement of classification with quantification, movement from two/three-point scale to five-point scale, etc. We participated in Subtask A of SESA-16, namely message polarity classification, a task that seeks to predict a sentiment label for some given text. We model the problem as a multi-class classifi- cation problem that combines the predictions given by two different classifiers: one is a Convolutional Neural Network (CNN) and the other is Logistic Re- gression (LR). The former takes embedding-based features while the latter utilizes various features such as lexicons, dictionaries, etc. The remainder of this paper is structured as fol- lows. In Section 2, we describe our system in de- tail, including feature description and approaches. In Section 3, we list the details of datasets for the ex- periments, along with hyperparameter settings and training techniques. In Section 4, we report the ex- periment results and present the corresponding dis- cussion. 2 System Description Our system aims at predicting the sentiment of a given message, i.e., whether the message expresses positive, negative or neutral emotion. To achieve that, we adopt two separate classifiers, CNN and LR, designed to utilize different types of features. The fi- nal prediction for sentiment is a combination of pre- dictions given by both classifiers. 2.1 Data Preprocessing Tweets often include informal text, making it es- sential to preprocess tweets before they are fed to the system. However, we keep the preprocessing to a minimum by only removing URLs and @User tags. We then further tokenize and tag tweets with arktweetnlp (Gimpel et al., 2011). In addition, all tweets are lower-cased. 2.2 Logistic Regression We use the LR classifier for features from sentiment lexicons and token clusters. We have used the fol- 139 Figure 1: CNN architecture for an example …

Content truncated to 500 words