The Trackback-Rank algorithm for the blog search

Abstract

Blog is a personal publishing tool which encourages users to contributions in the Web. As the number of blog entries and contributors (bloggers) grows at a very fast pace, they are increasingly filling the Web space. Thus effective search in the blogspace become more important. For effective search, the page ranking algorithm is one of the most critical techniques. Blogs have the structural features, which do not exist in the traditional Web, such as trackback links, tags, comments. For this reason, the page ranking algorithms for the traditional Web may not work effectively in the blogspace. In this paper, we propose a new ldquotrackback-rankrdquo algorithm which considers the features of blogs for more effective blog search. We evaluate bloggers' authority, trackback connectivity, and users' reactivity in order to rank blog entries. These factors are created and modified by the interaction among blog users. The blog users read and evaluate contents of blog entries and then interaction other users. Thereby, these factors implicitly reflect the contents quality of the entries, and the trackback-rank algorithm could improve the relevance of the search result to the queries. Our experiments on a collection of 62,906 blog entries shows that the trackback-rank algorithm can more effectively find relevant information compared to the traditional ranking algorithm.