Semi-supervised Stance Detection of Tweets Via Distant Network Supervision
Authors:
Subhabrata Dutta,
Samiya Caur,
Soumen Chakrabarti,
Tanmoy Chakraborty
Abstract:
Detecting and labeling stance in social media text is strongly motivated by hate speech detection, poll prediction, engagement forecasting, and concerted propaganda detection. Today's best neural stance detectors need large volumes of training data, which is difficult to curate given the fast-changing landscape of social media text and issues on which users opine. Homophily properties over the soc…
▽ More
Detecting and labeling stance in social media text is strongly motivated by hate speech detection, poll prediction, engagement forecasting, and concerted propaganda detection. Today's best neural stance detectors need large volumes of training data, which is difficult to curate given the fast-changing landscape of social media text and issues on which users opine. Homophily properties over the social network provide strong signal of coarse-grained user-level stance. But semi-supervised approaches for tweet-level stance detection fail to properly leverage homophily. In light of this, We present SANDS, a new semi-supervised stance detector. SANDS starts from very few labeled tweets. It builds multiple deep feature views of tweets. It also uses a distant supervision signal from the social network to provide a surrogate loss signal to the component learners. We prepare two new tweet datasets comprising over 236,000 politically tinted tweets from two demographics (US and India) posted by over 87,000 users, their follower-followee graph, and over 8,000 tweets annotated by linguists. SANDS achieves a macro-F1 score of 0.55 (0.49) on US (India)-based datasets, outperforming 17 baselines (including variants of SANDS) substantially, particularly for minority stance labels and noisy text. Numerous ablation experiments on SANDS disentangle the dynamics of textual and network-propagated stance signals.
△ Less
Submitted 5 January, 2022; v1 submitted 3 January, 2022;
originally announced January 2022.
Influence of NaMo App on Twitter
Authors:
Shreya Sharma,
Samiya Caur,
Hitkul,
Ponnurangam Kumaraguru
Abstract:
Social media plays a crucial role in today's society. It results in paradigm changes in how people relate and communicate, convey and exchange ideas. Moreover, social media has evolved into critical knowledge networks for consumers and also affects decision-making. In elections, social media became an integral part of political campaigning to reach a greater audience and gather more support. The 2…
▽ More
Social media plays a crucial role in today's society. It results in paradigm changes in how people relate and communicate, convey and exchange ideas. Moreover, social media has evolved into critical knowledge networks for consumers and also affects decision-making. In elections, social media became an integral part of political campaigning to reach a greater audience and gather more support. The 2019 Lok Sabha election saw a massive spike in the usage of online social media platforms such as Twitter, Facebook, and WhatsApp; with every major political party launching its own organized social media campaigns. In 2014, Bhartiya Janta Party (BJP) took one step ahead in organizing the campaign by launching its app - NaMo App. We focus our research on Twitter and NaMo App during the 2019 Lok Sabha elections and CAA protests. Twitter is a platform where every individual can express their views and is not biased. In contrast, NaMo App is one of the first apps centered around a specific political party. It acted as a digital medium for BJP for organizing the political campaign to make people's opinion in their favour. This research aims to characterize the role of the NaMo App in a more traditional network as Twitter in shaping political discourse and studies the existence of an online echo chamber. We began by analyzing the amount and type of content shared using the NaMo App on Twitter. We performed content and network analysis for the existence of the echo chamber. We also applied Hawkes process to see the influence that NaMo App has on Twitter. Through this research, we can conclude that the users who share content using NaMo App, may be part of an online echo chamber and are likely to be BJP workers. We show the reach and influence that the NaMo App has on Twitter is significantly less, indicating its inability to break through the diverse audience and change the narrative on Twitter.
△ Less
Submitted 4 September, 2021;
originally announced September 2021.
