Detecting trend and bursty keywords using characteristics of Twitter stream data

Abstract

Twitter is a very popular online social networking and micro-blogging service that enables its users to post and share text-based messages called tweets. The numbers of active users and tweets generated daily are enormous and hence they, collectively, can give crucial clues to several interesting problems such as public opinion analysis and hot trend detection. Especially, to find out hot issues and trends from tweets, detection of popular keywords is very important. In this paper, we propose a new scheme for detecting trend and bursty keywords from Twitter stream data. Our scheme is very robust in that it can handle typical usages such as various abbreviations, minor typing errors and spacing errors that occur very frequently when writing tweets on various mobile devices. We implemented a prototype system and performed various experiments to show the effectiveness of our scheme.