NLPZZX at SemEval-2018 Task 1: Using Ensemble Method for Emotion and Sentiment Intensity Determination

Proceedings of the 12th International Workshop on Semantic Evaluation (SemEval-2018), pages 116–122 New Orleans, Louisiana, June 5–6, 2018. ©2018 Association for Computational Linguistics NLPZZX at SemEval-2018 Task 1: Using Ensemble Method for Emotion and Sentiment Intensity Determination Zhengxin Zhang, Qimin Zhou, Hao Wu School of Information Science and Engineering, Yunnan University Chenggong Campus, Kunming, P.R. China {zzxynu,zqmynu}@gmail.com, haowu@ynu.edu.cn Abstract In this paper, we put forward a system that competed at SemEval-2018 Task 1: “Affect in Tweets”. Our system uses a simple yet ef- fective ensemble method which combines sev- eral neural network components. We partici- pate in two subtasks for English tweets: EI-reg and V-reg. For two subtasks, different com- binations of neural components are examined. For EI-reg, our system achieves an accuracy of 0.727 in Pearson Correlation Coefficient (all instances) and an accuracy of 0.555 in Pear- son Correlation Coefficient (0.5-1). For V-reg, the achieved accuracy scores are respectively 0.835 and 0.670. 1 Introduction Sentiment analysis is a research area in the field of natural language processing. It aims to detect the sentiment expressed by the author of some form of textual data and many deep learning ap- proaches have been successfully exploited (Cam- bria, 2016). The goal of SemEval-2018 Task 1 “Affect in Tweets” is to automatically determine the intensity of emotions and intensity of senti- ment of the tweeters from their tweets (Moham- mad et al., 2018). All tweets fall into three lan- guages: English, Arabic and Spanish. We par- ticipate in two subtasks for English tweets: EI- reg and V-reg. For EI-reg, all English tweets are separated into four emotions, anger, fear, joy and sadness. Every emotion has train, dev and test datasets. This subtask determines the inten- sity which is a real-valued score between 0 and 1 of emotion that represents the mental state of the tweeter. The instances with higher scores cor- respond to a greater degree of emotion than in- stances with lower scores. For V-reg, all English tweets are divided into three datasets: train, dev and test datasets. It determines the intensity of sentiment or valence that best represents the men- tal state of the tweeter a real-valued score between 0 and 1. The instances with higher scores cor- respond to a greater degree of positive sentiment than instances with lower scores. Both the two subtasks are regression tasks. For these two subtasks, we have adopted sep- arate ensemble method with existing neural net- work components (Brueckner and Schulter, 2014; Kim, 2014; Li and Qian, 2016; Yang et al., 2017) (see Figure 1). We use BiLSTM-CNN com- ponent, BiLSTM-Attention component and Deep BiLSTM-Attention component with different em- beddings for simple ensemble. In these subtasks, our final model is just an average of scores pro- vided by what we select from these single neu- ral network components. Every emotion or va- lence employs different ensemble method, so there are several distinct ensemble methods in the two subtasks. Experimental results show that our pro- posed ensemble methods are simple yet effective. The remainder of …

Content truncated to 500 words