The aim of this thesis is to summarize the main information (subevents) of a (main)event, using real time information. In this paper I will do this by predicting the football rankings of a season based on event-tweets which were sent during the corresponding matches. The tweets will be analyzed based on the outburst of tweet-streams and the word distribution used in tweets. The final score of a match is predicted by determining the most frequent score in the last tweet-outburst as the end result. This approach extracts the tweets containing the final score with a recall of 0.48 and a precision of 0.63. Goalscorers are predicted by extracting the most frequent player-score combinations in tweet-outbursts and making a logic scoring progress out of it. This approach extracts the tweets containing the goalscorers with a recall of 0.56 and a precision of 0.94. With the total approach 100% of the final scores and 91% of the goalscorers are predicted correct.

The complete Bachelor Thesis can be downloaded here:
Thesis