Exploiting Twitter Data for Music Recommendation
The goal of this master’s thesis is to develop a Music Recommender System, which utilizes tweets about music published on the microblogging service Twitter.
Therefore, in the first step, the tweets have to be matched to a song and an artist. This can be achieved by exploiting the URLs found in the tweets, but also by using regular expressions and available music databases, if the tweet doesn’t contain a useful URL. In the second step diverse Collaborative Filtering approaches, Association Rules as well as Hybrid Methods are evaluated, which are then used for the actual music recommendation. Comparing it to other music recommendation services can serve as an assessment of the Recommender System, but also a user experiment might be conducted.