Document Type
Article
Publication Date
January 2018
Journal
The Canadian Journal of Information and Library Science
Volume
42
Issue
3-4
First Page
154
Last Page
175
Abstract
Automatic clickbait detection is a relatively novel task in natural language processing (NLP) and machine learning (ML). “Clickbait” is a hyperlink created primarily to attract attention to its target content. This article introduces a binary classifier, the Language and Information Technology Research Lab (LiT.RL, pronounced “literal”) Clickbait Detector, which automatically distinguishes clickbait from nonclickbait. We used NLP and ML for 38 textual features, contrasting clickbait with “headlinese.” When tested on 11,000 hyperlinks, it achieves 94 per cent accuracy using a support vector machine. Integrated with the LiT.RL News Verification Browser, a downloadable stand-alone research tool, the Clickbait Detector user interface shows automated real-time colour-coded analysis of any news website.
Included in
Computational Engineering Commons, Computational Linguistics Commons, Computer Engineering Commons, Library and Information Science Commons