Abstract: With proliferation of social networking sites, large number of tweets, posts and messages are available which can be analysed to extract some useful information. Identifying topic experts from these posts is a daunting task. Collecting and storing various types twitter data in terms of real world dataset is difficult and analyzing of data is too tedious. Earlier approaches for expert finding partially utilized relations among user and twitter list. In this project, a semi-supervised graph based ranking (SSGR) solution has been implemented to compute the global authority of users in the offline mode. The local relevance between users and the given query is computed in the online mode. By taking the advantage of the local relevance and global authority of users, all the users are ranked and the top users with highest ranking are selected.

Keywords: twitter, expert search, micro-blogging, list, and graph based ranking.