ISCLAB at SemEval-2018 Task 1: UIR-Miner for Affect in Tweets

Proceedings of the 12th International Workshop on Semantic Evaluation (SemEval-2018), pages 286–290 New Orleans, Louisiana, June 5–6, 2018. ©2018 Association for Computational Linguistics 1 ISCLAB at SemEval-2018 Task 1: UIR-Miner for Affect in Tweets Meng Li1, 2, Zhenyuan Dong1, Zhihao Fan1, Kongming Meng1, Jinghua Cao1, Guanqi Ding1, Yuhan Liu1, Jiawei Shan1, Binyang Li1* 1 School of Information Science and Technology, University of International Relations 2 University of Pittsburgh * Corresponding author mel165@pitt.edu; {byli, zydong, zhfan, kmmeng, jhcao, gqding, jwshan}@uir.edu.cn Abstract This paper presents a UIR-Miner system for emotion and sentiment analysis evaluation in Twitter in SemEval 2018. Our system con- sists of three main modules: preprocessing module, stacking module to solve the in- tensity prediction of emotion and senti- ment, LSTM network module to solve multi-label classification, and the hierar- chical attention network module for solv- ing emotion and sentiment classification problem. According to the metrics of SemEval 2018, our system gets the final scores of 0.636, 0.531, 0.731, 0.708, and 0.408 in terms of Pearson Correlation on 5 subtasks, respectively. 1 Introduction Recently, social media platforms are becoming more and more popular, such as Twitter mi- croblogging, Facebook, and so on. Through these platforms, online users would like to share their opinions and emotions. Therefore, the analysis about the information on “affect” in the social me- dia has attracted much interest from both academia and industries. However, the short texts are usually consisted of informal expressions with much casual forms and emoticons, it brings great challenges for such re- search. For this purpose, SemEval organized the evalu- ation of sentiment analysis on Tweet. This year comes the fifth edition that consists of new genres, including emotion intensity regression task, emo- tion intensity ordinal classification task, sentiment intensity regression task, sentiment degree ordinal classification task, and emotion classification task (Mohammad et al., 2018). Figure 1: System architecture. We participated in SemEval 2018 task 1 for Eng- lish, i.e. Affect in Tweet. Our system considers EI- reg and V-reg (subtask A and C) as regression prob- lems to get the emotion intensity and sentiment in- tensity by using regression models, while regards EI-oc and V-oc (subtask B and D) as categorization problems to classify each tweet into its correspond- ing emotion category and sentiment category by implementing hierarchical attention networks. Moreover, subtask E, i.e., E-c, is considered as a multi-label classification task. This paper is organized as follows. Section 2 overviews the framework of our system. Section 3 describes the methods for subtask A and C. Section 4 describes the hierarchical attention networks for subtask B and D. Subtask E will be introduced in Section 5. Section 6 presents the evaluation results. Section 7 will conclude this paper. 2 System Overview The architecture of UIR-Miner is shown in Figure 1. UIR-Miner system is comprised of 4 modules: (1) Preprocessing module: involves data cleaning, topic classification, and tweets embedding. 286 2 (2) Regressor module: creates an ensemble regres- sor model by using different basic models sim- ultaneously to calculate the emotion intensity …

Content truncated to 500 words