Classification of spammer and nonspammer content in online social network using genetic algorithm-based feature selection. Issue 5 (27th May 2020)

• Record Type: Journal Article
• Authors: Sahoo, Somya Ranjan; Gupta, B. B.
• Is Part Of: Enterprise information systems. Volume 14:Issue 5(2020)
• Page Start: 710
• Page End: 736
• Publication Date: 2020-05-27
• Subjects: Online social networking -- genetic algorithm -- spammer -- machine learning; Information storage and retrieval systems -- Periodicals; Management information systems -- Periodicals; Electronic commerce -- Periodicals; 658.4038011
• DOI: 10.1080/17517575.2020.1712742
• Languages: English
• ISSNs: 1751-7575

Abstract

ABSTRACT: The emergence of online social network invokes social actors to share their personal information digitally. Moreover, it provides the facility to maintain their links with people of same interest globally. Take advantage of these services; it has become a fascinating testbed to invite various threats like a spammer. Detection of spammer in OSN is one of the most critical tasks. Spammer not only spreads unwanted or bad advertisement but does certain malicious activity in others' profiles. By clearly understanding the activities of different threats, some incremental and accurate approaches are needed for detecting spammer content and profiles involved in these activities by using social network services. Therefore, the focus of this article is to detect spammer content and account, specifically on the leading microblogging platform called Twitter. We propose a hybrid approach which leverages the capabilities of various machine learning algorithms to separate spammer and nonspammer contents and account. Initially, the optimisation algorithm called genetic algorithm analyses the various features and selects the best suitable features that influence the behaviour of user account, and these features are then used to train classifiers. Our framework achieved to severalise spammer and nonspammer content in an effective way. Finally, to prove the efficiency of our proposed framework, a comparative analysis is conducted with some existing state-of-art techniques. The experimental analysis shows that our approach achieves a high detection rate of 99.6%, which is better than other state-of-art techniques.