Multi-feature based similarity among entries on media portals

Abstract

Similar measures play an important role in information processing and have been widely investigated in computer science. With the exploration of social media such as Youtube, Wikipedia, Facebook etc., a huge number of entries have been posted on these portals. They are often described by means of short text or sets of words. Discovering similar entries based on such texts has become challenges in constructing information searching or filtering engines and attracted several research interests. In this paper, we firstly introduce a model of entries posted on media or entertainment portals, which is based on their features composed of title, category, tags, and content. Then, we present a novel similar measure among entries that incorporates their features. The experimental results show the superiority of our incorporation similarity measure compared with the other ones.