Search using social networks

Abstract

In this thesis, we present an approach to the problem of personalized web search which makes use of the searcher's social network, in addition to the hyper-link based score used in most search engines. This combination of social and absolute search scores aims to improve the visibility of information that is relevant to the searcher, while maintaining any absolute measures of document importance . In our approach we adopt a flexible framework for combining information from different sources using Rank Aggregation techniques. Our search system, implemented using Java and Python, covers all the events and web pages present on MIT owned websites. We discuss the theory, design,and implementation of this system in details.