| In online social networks, users create a profile by setting some attributes that help in characterizing the user. We call these profile attributes the keywords of the user. In this work, we focus on social networks based on keywords and study how keywords can be effectively used to model and search in such networks. We make two main contributions in this thesis.First, we study the relationship between semantic similarity of user keywords and the social network topology. We present a 'forest' model to categorize keywords and define similarity functions between a pair of users. Based on that, we model the social network topology and validate the model against a real life social network graph.Second, we study unstructured keyword based social networks and propose an information flow model to disseminate keywords. Then, based on the given information flow, we design and develop a search algorithm to find users who have keywords, present in the search query, as part of their profile attributes. The search algorithm is based on a linear combination of topological distance and trust metrics. We observe an improvement in orders of magnitude when the search algorithm is compared to breadth first search. |
