Clustering and Ranking for Web Information Retrieval

This book investigates several research problems which arise in modern Web Information Retrieval. First of all we consider the fact that there are many situations where a flat list of ten search results are not enough, and that the users might desire to have a larger number of results grouped on-the-fly in folders of similar topics. In this book, we describe Snaket, a hierarchical clustering meta-search engine which personalizes searches according to the clusters selected on-the-fly by users. Second, we consider those situations where users might desire to access fresh information such as news articles. We present a new ranking algorithm suitable for ranking those fresh type of information. Third, we will discuss numerical methodologies for accelerating the ranking methodologies used in Web Search. An important achievement for this book is that we show how to address the above predominant issues of Web Information Retrieval by using clustering and ranking methodologies. We demonstrate that both clustering and ranking have a mutual reinforcement property that has not yet been studied intensively.