AndrejJan at SemEval-2019 Task 7: A Fusion Approach for Exploring the Key Factors pertaining to Rumour Analysis

Proceedings of the 13th International Workshop on Semantic Evaluation (SemEval-2019), pages 1083–1089 Minneapolis, Minnesota, USA, June 6–7, 2019. ©2019 Association for Computational Linguistics 1083 AndrejJan at SemEval-2019 Task 7: A Fusion Approach for Exploring the Key Factors pertaining to Rumour Analysis Andrej Janchevski, Sonja Gievska Faculty of Computer Science and Engineering Ss. Cyril and Methodius University Rugjer Boshkovikj 16, Skopje, Republic of North Macedonia andrej.jancevski@students.finki.ukim.mk sonja.gievska@finki.ukim.mk Abstract The viral spread of false, unverified and mis- leading information on the Internet has at- tracted a heightened attention of an inter- disciplinary research community on the phe- nomenon. This paper contributes to the re- search efforts of automatically determining the veracity of rumourous tweets and classifying their replies according to stance. Our research objective was to investigate the interplay be- tween a number of phenomenological and con- textual features of rumours, in particular, we explore the extent to which network struc- tural characteristics, metadata and user pro- files could complement the linguistic analysis of the written content for the task at hand. The current findings strongly demonstrate that sup- plementary sources of information play signif- icant role in classifying the veracity and the stance of Twitter interactions deemed to be ru- mourous. 1 Introduction Social networks continue to contribute to the way people connect, stay informed and contribute above and beyond what concerns their lives. How- ever, these platforms currently play a crucial role in viral spreading of false information, and no doubt will become even more instrumental in the future. Consequently, it will be increasingly important to devise systems and establish prac- tices of automatically identifying, filtering and la- beling false information in order to help people make sense of information dispersed through so- cial channels. The damaging consequences of malicious in- tents to misinform, confuse and provoke through social media platforms remain persistently high in the discussions of researchers and practition- ers. Perhaps even more so than the benefits of information dissemination. While the veracity of some false information can be determined unam- biguously from external sources, it is a major chal- lenge to verify the truthfulness of rumourous post- ings. Indeed, predictive analysis remains the pri- mary manner in which social platforms could face the challenge of identifying rumours and taking appropriate actions. Over the past several years, research has emerged and at the same time, many challenges remain. Decoding elusive social phenomenon such as spreading rumours is challenging not only due to their complexity, but also due to the diver- sity of the underlying causes and heterogeneity in their manifestations. Rumours represent in- tertwined threads of sensemaking Bordia and Di- fonzo (2004) that are initiated and spread by peo- ple trying to explain, solve and remove uncertainty relating to events and persons that attract public interest Peterson and Gist (1951). We argue that solutions that address these multi-dimensional is- sues cannot solely rely on natural language pro- cessing (NLP). Combining natural language pro- cessing with social analytics extends beyond the traditional realms of either technology to a variety …

Content truncated to 500 words