Hashtag Sense Clustering Based on Temporal Similarity

Hashtag Sense Clustering Based on Temporal Similarity Giovanni Stilo∗ University of Rome Sapienza Paola Velardi∗ University of Rome Sapienza Hashtags are creative labels used in micro-blogs to characterize the topic of a message/discussion. Regardless of the use for which they were originally intended, hashtags cannot be used as a means to cluster messages with similar content. First, because hashtags are created in a spontaneous and highly dynamic way by users in multiple languages, the same topic can be associated with different hashtags, and conversely, the same hashtag may refer to different topics in different time periods. Second, contrary to common words, hashtag disambiguation is complicated by the fact that no sense catalogs (e.g., Wikipedia or WordNet) are available; and, furthermore, hashtag labels are difficult to analyze, as they often consist of acronyms, concatenated words, and so forth. A common way to determine the meaning of hashtags has been to analyze their context, but, as we have just pointed out, hashtags can have multiple and variable meanings. In this article, we propose a temporal sense clustering algorithm based on the idea that semantically related hashtags have similar and synchronous usage patterns. 1. Introduction Hashtags are frequently used in social networks to tag message content, although their intended use may differ in different networks. Unlike other tagging systems (e.g., Flickr1), hashtags in Twitter are used more as a symbol of community mem- bership than for organizing a message content, as remarked in Huang, Thornton, and Efthimiadis (2010). Because interest groups in Twitter are very dynamic, hash- tag popularity surges and decays; and, furthermore, the same hashtag might refer to different events in different time periods. For example, recently Jawbone tried a #knowyourself campaign on Instagram,2 only to find that the hashtag was already being used generically by thousands of users in all sorts of different contexts. In addition to polysemy, there is also a problem of synonymy: Because new hashtags are freely and continuously introduced by the users, different hashtags, for example ∗ Dipartimento di Informatica, University of Rome Sapienza, Via Salaria 113, Rome, Italy. E-mail: {stilo,velardi}@di.uniroma1.it. 1 https://www.flickr.com/. 2 http://tinyurl.com/l5w6pt6. Submission received: 7 March 2015; revised version received: 18 April 2016; accepted for publication: 14 June 2016. doi:10.1162/COLI a 00277 © 2017 Association for Computational Linguistics Published under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0) license Computational Linguistics Volume 43, Number 1 [#FinaleCDM2014, #WCFinal2014], [#RussiainvadedUkraine, #UkraineUnderAttack] may share the same meaning, sometimes also because of multilinguality. Another prob- lem regarding hashtags—again, not found in other content tagging policies in social media—is obscurity. Hashtags are often hard to interpret for both humans and ma- chines, as they may consist of acronyms [#MH17, #GERBRA, #RIP], concatenated words [#VamosArgentinaQueSePuede], neologisms (#procrastitweet), or abbreviations (#tcot3). Together, these problems reduce the effectiveness of Twitter hashtags as a means both for tracing users’ interests (owning to obscurity and sense shifts), and for capturing the worldwide impact of trending topics (because of synonymy and multilinguality). Despite this, however, better methods for analyzing the semantics of hashtags …

Content truncated to 500 words