Trendnet algorithm: Towards personalization of twitter trends

Abstract

Twitter displays a list of currently popular keywords and hashtags on the user homepage which are referred as trends. Trends play an important role in discovering the hottest emerging topics of discussion and also help in categorizing the tweets through which the user can easily find similar tweets in that group. Twitter provides its user with a list of top ten trending topics but these trends are general topics which are popular based on the user location and are not context sensitive. These suggestions are not personalized. This paper examines an application for finding personalized trending topics on Twitter. We propose a novel real time trend recommendation system referred as TrendNet that helps its users to find what is currently popular in their network of friends by considering both the tweet content and the social structure. Comprehensive experiments on real Twitter users having different interests were conducted in order to evaluate the effectiveness of the algorithm. The results demonstrate that our scheme provides more accurate and personalized recommendations of trends as compared to the existing scheme.