A deep model based on Lure and Similarity for Adaptive Clickbait Detection

Abstract

With the rapid development of the Internet and the intensified competition in the media, media professionals or self-media editors often attract attentions and clicks by clickbait news to enlarge the revenue. Clickbait news usually have two essential connotations: luring headline and low similarity between headline and target content. Due to the huge amount and dynamics of online news, automated clickbait detection methods are indispensable. Traditional machine learning-based-methods suffered from heavy feature engineering and poor performance. Some deep learning-based approaches considered only part of the news content or directly fuse all the information. In this paper, by considering the human recognition behavior for clickbait, i.e. the final decision of whether the news is clickbait depends on both the luring degree of headlines and the similarity between the headlines and the target descriptions, we propose a novel integrative and adaptive method based on Lure and Similarity for Adaptive Clickbait Detection (LSACD). In the LSACD11A Deep Model based on Lure and Similarity for Adaptive Clickbait Detection. architecture, a deep model by combining lure and similarity is designed, and an adaptive prediction mechanism is utilized to make reasonable prediction with the help of different contributions of both modules. We construct a novel and real Chinese clickbait dataset which contains nearly 5000 news with high click rate. Experiments on open Twitter news dataset and the Chinese clickbait dataset show the superiority, effectiveness and reliability of our proposed LSACD model.