Multi-Task Stance Detection with Sentiment and Stance Lexicons

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 6299–6305, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 6299 Multi-Task Stance Detection with Sentiment and Stance Lexicons Yingjie Li Computer Science University of Illinois at Chicago Chicago, IL, USA yli300@uic.edu Cornelia Caragea Computer Science University of Illinois at Chicago Chicago, IL, USA cornelia@uic.edu Abstract Stance detection aims to detect whether the opinion holder is in support of or against a given target. Recent works show improve- ments in stance detection by using either the attention mechanism or sentiment informa- tion. In this paper, we propose a multi-task framework that incorporates target-specific at- tention mechanism and at the same time takes sentiment classification as an auxiliary task. Moreover, we used a sentiment lexicon and constructed a stance lexicon to provide guid- ance for the attention layer. Experimental re- sults show that the proposed model signifi- cantly outperforms state-of-the-art deep learn- ing methods on the SemEval-2016 dataset. 1 Introduction With the rapid growth of social media, user opin- ions towards various targets, e.g., politicians and religion, are abundant. These opinions can help optimize management systems and can gain in- sight into important events, e.g., presidential elec- tions. The stance detection task aims to determine whether people are in favor of, against, or neu- tral towards a specific target. This task is simi- lar to the three-way aspect-level sentiment analy- sis that determines sentiment polarity towards as- pect terms. However, different from aspect-level sentiment analysis, the target in stance detection might not be explicitly mentioned in a given sen- tence. Consider the following example tweet: @realDonaldTrump is the only honest voice of the @GOP and that should scare the shit out of ev- eryone! #SemST. Target: Hillary Clinton; Stance: Against; Sentiment: Positive. Observe that even though target Hillary Clinton does not appear in this tweet, we can still infer that opinion holder is less likely to be in favor of Hillary Clinton. There- fore, identifying target related information is of vi- tal importance to stance detection. Previous studies to stance detection used feature engineering (Mohammad et al., 2016b), Convolu- tional Neural Networks (CNNs) (Vijayaraghavan et al., 2016; Wei et al., 2016) and Recurrent Neu- ral Networks (RNNs) (Zarrella and Marsh, 2016). However, they failed to take target information into considerations. In order to address this issue, several target-specific attention mechanisms (Du et al., 2017; Zhou et al., 2017; Sun et al., 2018) have been proposed to embed target information into sentence representations. Sentiment information has also been proven to be beneficial to stance detection. For example, Sobhani et al. (2016) found that sentiment lexi- con features are useful for stance detection when combined with other features. Instead of directly integrating sentiment features into vector repre- sentations, Sun et al. (2018) proposed a hierarchi- cal attention network that learns the importance of sentiment information automatically. Later, Sun et al. (2019) proposed a joint model that deter- mines …

Content truncated to 500 words