Target-Aware Data Augmentation for Stance Detection

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 1850–1860 June 6–11, 2021. ©2021 Association for Computational Linguistics 1850 Target-Aware Data Augmentation for Stance Detection Yingjie Li University of Illinois at Chicago yli300@uic.edu Cornelia Caragea University of Illinois at Chicago cornelia@uic.edu Abstract The goal of stance detection is to identify whether the author of a text is in favor of, neu- tral or against a specific target. Despite sub- stantial progress on this task, one of the re- maining challenges is the scarcity of annota- tions. Data augmentation is commonly used to address annotation scarcity by generating more training samples. However, the aug- mented sentences that are generated by exist- ing methods are either less diversified or in- consistent with the given target and stance la- bel. In this paper, we formulate the data aug- mentation of stance detection as a conditional masked language modeling task and augment the dataset by predicting the masked word con- ditioned on both its context and the auxiliary sentence that contains target and label informa- tion. Moreover, we propose another simple yet effective method that generates target-aware sentence by replacing a target mention with the other. Experimental results show that our pro- posed methods significantly outperforms pre- vious augmentation methods on 11 targets. 1 Introduction Nowadays, people often take to social media to express their stances toward specific targets (e.g., political figures or abortion). These stances in an aggregate can provide valuable information for ob- taining insight into some important events such as presidential elections. The goal of the stance de- tection task is to determine from a piece of text whether the author of the text is in favor of, neu- tral or against toward a specific target (Mohammad et al., 2016; Lin et al., 2019), which indicates that all elements, the sentence, the target, and the la- bel, are used to train a stance detection model. We can further classify the task as single-target stance detection and multi-target stance detection (Küçük and Can, 2020; AlDayel and Magdy, 2020) where we need to detect the stances toward two different targets simultaneously. Orig. We all have a duty to protect the sanctity of life. G1 We all have a life to protect the sanctity of duty. � G2 We all have a duty to protect the sanctitude of life. � G3 We all have a responsibility to protect the unborn lives. � Table 1: Examples of data augmentation on the target “Legalization of Abortion”. One of the biggest challenges for stance detec- tion tasks is the scarcity of annotated data. Data augmentation (DA) is an effective strategy for han- dling scarce data situations. However, we face three main obstacles when applying the existing augmentation methods to the stance detection tasks. First, existing augmentation methods do not gener- alize well, which means some methods are tailored to specific tasks and models, and thus difficult to be extended to the stance detection tasks. Second, con- …

Content truncated to 500 words