Leveraging Topic Extraction for Hashtag Recommendations

The microblogging platform Twitter provides its users with the use of so-called hashtags which enable the users to categorize tweets and hence, group tweets related to a certain topic. Hashtags may easily be added to the tweet itself by stating the tag preceded by a #-sign, like e.g. in #tourdefrance. The DBIS group already implemented a prototype for the recommendation of hashtags based on collaborative filtering approaches. This approach basically recommends hashtags which have been used by syntactically similar tweets.

The goal of this thesis is to investigate whether the extraction of the topic of the tweet from either the tweet itself or from URLs contained in the tweet may contribute to the recommendation hashtags. The analysis of different topic extraction methods, the implementation of a prototype and an according evaluation are part of this thesis.