Finding Topic Experts in the Twitter Dataset Using LDA Algorithm

Abstract

In microblogging services like Twitter, the expert judgment problem has gained increasing attention in social media. Twitter is a new type of social media that provides a publicly available way for users to publish 140-character short messages (tweets). However, previous methods cannot be directly applied to twitter experts finding problems. They generally rely on the assumption that all the documents associated with the candidate experts contain tacit knowledge related to the expertise of individuals. Whereas it might not be directly associated with their expertise, i.e., who is not an expert, but may publish/re-tweet a substantial number of tweets containing the topic words. Recently, several attempts use the relations among users and twitter list for expert finding. Nevertheless, these strategies only partially utilize such relations. To address these issues a probabilistic method is developed to jointly exploit three types of relations (i.e., follower relation, user-list relation and list-list relation) for finding experts. LDA algorithms are used for finding topic experts. LDA is based upon the concept of searching for a linear combination of variables (predictors) that best separates two classes (targets). Semi-supervised graph-based ranking approach (SSGR) to offline calculate the global authority of users. Then, the local relevance between users and the given query is computed. Then, the rank of all the users is found and the top-N users with the highest-ranking scores. Therefore, the proposed approach can jointly exploit the different types of relations among users and lists for improving the accuracy of finding experts on a given topic on Twitter.