Weakly Supervised Tweet Stance Classification by Relational Bootstrapping

Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 1012–1017, Austin, Texas, November 1-5, 2016. c⃝2016 Association for Computational Linguistics Weakly Supervised Tweet Stance Classification by Relational Bootstrapping Javid Ebrahimi and Dejing Dou and Daniel Lowd Department of Computer and Information Science, University of Oregon Eugene, Oregon 97403, USA {javid,dou,lowd}@cs.uoregon.edu Abstract Supervised stance classification, in such do- mains as Congressional debates and online forums, has been a topic of interest in the past decade. Approaches have evolved from text classification to structured output predic- tion, including collective classification and se- quence labeling. In this work, we investigate collective classification of stances on Twitter, using hinge-loss Markov random fields (HL- MRFs). Given the graph of all posts, users, and their relationships, we constrain the pre- dicted post labels and latent user labels to cor- respond with the network structure. We focus on a weakly supervised setting, in which only a small set of hashtags or phrases is labeled. Using our relational approach, we are able to go beyond the stance-indicative patterns and harvest more stance-indicative tweets, which can also be used to train any linear text classi- fier when the network structure is not available or is costly. 1 Introduction Stance classification is the task of determining from text whether the author of the text is in favor of, against, or neutral towards a target of interest. This is an interesting task to study on social networks due to the abundance of personalized and opinion- ated language. Studying stance classification can be beneficial in identifying electoral issues and under- standing how public stance is shaped (Mohammad et al., 2015). Twitter provides a wealth of information: pub- lic tweets by individuals, their profile information, whom they follow, and more. Exploiting all these pieces of information, in addition to the text, could help build better NLP systems. Examples of this approach include user preference modeling (Li et al., 2014), stance classification (Rajadesingan and Liu, 2014), and geolocation identification (Jurgens, 2013; Rahimi et al., 2015). For stance classification, knowing the author’s past posting behavior, or her friends’ stances on issues, could improve the stance classifier. These are inherently structured problems, and they demand structured solutions, such as Statis- tical Relational Learning (SRL) (Getoor, 2007). In this paper, we use hinge-loss Markov random fields (HL-MRFs) (Bach et al., 2015), a recent develop- ment in the SRL community. SemEval 2016 Task 6 organizers (Mohammad et al., 2016) released a dataset with Donald Trump as the target, without stance annotation. The goal of the task was to evaluate stance classification sys- tems, which used minimal labeling on phrases. This scenario is becoming more and more relevant due to the vast amount of data and ever-changing nature of the language on social media. This is critical in ap- plications in which a timely detection is highly de- sired, such as violence detection (Cano Basave et al., 2013) and disaster situations. Our work is the first to use SRL for stance classi- fication on Twitter. …

Content truncated to 500 words