Comprehensive Survey on Techniques of Topic Evolution Mining

With the tremendous increase in use of social media, information is constantly propagating online. Certain topics or events are popular at a given time, and are said to be trending on social networks such as twitter. Given the ease and pace at which information is generated on twitter, the trending topics are constantly evolving. When a twitter user starts tweeting with respect to the root event, they give way to multiple sub-events and subtopics, which themselves continue to evolve. Evolution of topics in such highly co related topics is quite interesting. Mining the evolution of such topics is an intriguing task which can yield interesting facts about the socio-political, economical or cultural climate of a place or community. It has applications in marketing, political campaigns, trend analysis and so on. Such research can also enable the tracking and prediction of twitter activity in the future. Due to the massive data, it has become increasingly difficult to cluster the interesting events from vast social media data. This paper comprehensively describes and summarizes the previous research in topic evolution of twitter data and provides a comparative analysis of the works in this domain, highlighting areas of future improvement.