Contrastive Language Adaptation for Cross-Lingual Stance Detection

Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing, pages 4442–4452, Hong Kong, China, November 3–7, 2019. c⃝2019 Association for Computational Linguistics 4442 Contrastive Language Adaptation for Cross-Lingual Stance Detection Mitra Mohtarami1, James Glass1, Preslav Nakov2 1MIT Computer Science and Artificial Intelligence Laboratory, Cambridge, MA, USA 2Qatar Computing Research Institute, HBKU, Doha, Qatar {mitra,glass}@csail.mit.edu; pnakov@hbku.edu.qa Abstract We study cross-lingual stance detection, which aims to leverage labeled data in one language to identify the relative perspective (or stance) of a given document with respect to a claim in a different target language. In particular, we introduce a novel contrastive language adap- tation approach applied to memory networks, which ensures accurate alignment of stances in the source and target languages, and can ef- fectively deal with the challenge of limited la- beled data in the target language. The evalua- tion results on public benchmark datasets and comparison against current state-of-the-art ap- proaches demonstrate the effectiveness of our approach. 1 Introduction The rise of social media has enabled the phe- nomenon of “fake news,” which could target spe- cific individuals and can be used for deceptive pur- poses (Lazer et al., 2018; Vosoughi et al., 2018). As manual fact-checking is a time-consuming and tedious process, computational approaches have been proposed as a possible alternative (Popat et al., 2017; Wang, 2017; Mihaylova et al., 2018, 2019), based on information sources such as social me- dia (Ma et al., 2017), Wikipedia (Thorne et al., 2018), and knowledge bases (Huynh and Papotti, 2018). Fact-checking is a multi-step process (Vla- chos and Riedel, 2014): (i) checking the reliability of media sources, (ii) retrieving potentially relevant documents from reliable sources as evidence for each target claim, (iii) predicting the stance of each document with respect to the target claim, and fi- nally (iv) making a decision based on the stances from (iii) for all documents from (ii). Here, we focus on stance detection which aims to identify the relative perspective of a document with respect to a claim, typically modeled using labels such as agree, disagree, discuss, and unrelated. Current approaches to stance detection (Bar-Haim et al., 2017; Dungs et al., 2018; Kochkina et al., 2018; Sobhani et al., 2017; Mohtarami et al., 2018) are well-studied in mono-lingual settings, in partic- ular for English, but less attention has been paid to other languages and cross-lingual settings. This is partially due to domain differences and to the lack of training data in other languages. We aim to bridge this gap by proposing a cross-lingual model for stance detection. Our model leverages resources of a source language (e.g., English) to train a model for a target lan- guage (e.g., Arabic). Furthermore, we propose a novel contrastive language adaptation approach that effectively aligns samples with similar or dissimilar stances across source and target lan- guages using task-specific loss functions. We ap- ply our language adaptation approach to memory networks (Sukhbaatar et al., 2015), which have been found effective …

Content truncated to 500 words