Finding people and their utterances in social media

At the beginning of the twenty-first century, the web entered a phase of explosive growth. A multitude of platforms has become available for users to publish information, communicate with others, connect to like-minded individuals, and share anything that they wanted to share. These platforms are commonly known as social media. Social media enables many-to-many-communication: many people can create content, which in turn can, in principle, by read by many others. However, to make interesting content findable for many people, machines need to be able to identify the right pieces of content or the appropriate content creators. That is, we need ways to intelligently access information in social media. The main motivation for the research in this thesis is that we want to enable intelligent access to, and analysis of, the information contained in social media. To this end, we must determine the topical relevance of social media documents, while countering the specific challenges posed by their noisy nature. We identify two entry points for accessing information in social media: (i) the people creating the social media and (ii) their individual utterances. These entry points act as a doorway to the information in social media. The main aim of the thesis is to improve searching for people and their utterances in social media, thereby offering intelligent access to the information it contains. In this thesis we explore ways to improve retrieval effectiveness, for both people and their utterances. Methods range from analysis of a people search engine to credibility-inspired indicators in blogs, from blogger finding to using real-world context in query modeling. We gain new insights into search behavior and show that our methods help improve retrieval effectiveness for multiple information access tasks.