Analysis and Prediction of Dyads in Twitter
| Autor: | Yannis Korkontzelos, Isa Inuwa-Dutse, Mark Liptrott |
|---|---|
| Jazyk: | angličtina |
| Předmět: |
Sociological theory
business.industry Computer science Deep learning Closeness 02 engineering and technology Data science Trustworthiness 020204 information systems User group 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Artificial intelligence Macro Cluster analysis business Dyad |
| Zdroj: | Lecture Notes in Computer Science Lecture Notes in Computer Science-Natural Language Processing and Information Systems Natural Language Processing and Information Systems-24th International Conference on Applications of Natural Language to Information Systems, NLDB 2019, Salford, UK, June 26–28, 2019, Proceedings Natural Language Processing and Information Systems ISBN: 9783030232801 NLDB |
| ISSN: | 0302-9743 1611-3349 |
| DOI: | 10.1007/978-3-030-23281-8_25 |
| Popis: | Social networks are useful for linking micro and macro levels of sociological theory by enabling the analysis of various forms of relationships. In social science, a taxonomy of social relationships is described as a function of closeness among users. The closer the users are, the more cohesive and trustworthy. Identifying dyadic ties, pairs of fully connected users, on Twitter is challenging due to the flexible and eccentric underlying connection patterns. The ability to follow anyone results in many unidirectional connections between socially disconnected users and ultimately affects clustering users and, in turn, the veracity of online content. Major challenges towards effective user clustering are the low number of dyads and efficient methods to identify more. In this study, we query over 17M verified and unverified Twitter user accounts and retrieve dyadic ties. In the collected data, \(55\%\) and \(21\%\) of unverified and verified profiles, respectively, participate in dyadic ties. We describe the importance of dyads in the detection of cohesive user groups and how they may be used to validate trustworthiness. We demonstrate how identifying and using dyadic ties will improve Twitter analysis, in the future. Finally, we develop a deep learning model for dyad prediction. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|