AAAI Publications, Fifth International AAAI Conference on Weblogs and Social Media

Font Size: 
Describing the Web in less than 140 Characters
Stéphane Raux, Nils Grünwald, Christophe Prieur

Last modified: 2011-07-05

Abstract


Links curation, that is, finding relevant information within the World-Wide-Web and its ever-growing amount of content is a crucial problem for information access. Hyperlinks recommendation has been for a long time a common way to share references between web users, be it by e-mail exchanges, instant messages or forums. We explore in this paper how Social Media extend this recommendation practice by focusing on the citation of hyperlinks on Twitter. We investigate how people deal with the strong limitation of 140 characters per message, showing that this constraint encourages people to perform a good synthesis of the content they are linking to. We take advantage of this practice to efficiently cluster the actual content of the linked pages with an algorithm based on lexical proximities between messages. Our method yields topical clusters that are consistent with the dynamics of user interests with no need to extract text from the pages themselves.

Full Text: PDF