First and foremost, I would like to thank Prof. Dr. François Bry for offering this thesis and the advice he provided. Besides, I would like to thank Alexander Pohl for his guidance during the development of my work and especially for his patience with me whenever I got frustrated. He was always a good listener and answered my questions promptly. I want to express my eternal gratitude to my parents for their support and love. Finally, I would like to thank Johann Kratzer, my eternal love and best friend. Completing this work would have been more difficult without his support and humour. i Erklärung Hiermit versichere ich, dass ich die vorliegende Arbeit selbständig verfasst habe und keine anderen als die angegebenen Hilfsmittel verwendet habe. München, den 25. Januar 2012

We investigate in this paper information retrieval in microblogs using features. Microbloggers, besides posting microblogs, search for fresh and relevant information related to their interests. The majority of approaches that collect information from microblogs use features sush as the recency of the microblog, the authority of his/her author..., to narrow the quality of their results. In this paper, we evaluated state-of-the-art features to determine those that best reflect relevance. Then, we used the selected features to learn models to determine their effectiveness in a microblogs search task. We conducted experiments using the dataset and topics given in TREC Mircroblogs 2011 and 2012 tracks. Experiments show that content, hypertextuality and recency are the best predictors of relevance. We also found that Naive Bayes was the most effective learning approach for this type of classification.