Foxinfo1.0: A Chinese Topic-Oriented Search Engine

Abstract

Topic-oriented search engine (topic-search) is a new IR service which provides compounded types of information with certain user queried topic in one page. It firstly categorizes user query into a certain domain, and then organizes several types of information based on the query keywords into a magazine-style topic page for user. In this paper, we propose a Chinese topic-oriented search engine service, named as Foxinfo1.0, which provides a 360 degree view of the topic that interests the users who're seeking it. Different from the original topic-search which employs the keyword-based topic to retrieve and aggregate relevant information, Foxinfo1.0 could organize information from different abstraction level by employing tag to describe the topics of queries and information. Further, in order to predict tags from queries and web documents, a tag prediction algorithm named as CTAG is proposed, which could concern tags from different levels, and performs much better than the baseline method like AutoTag.