Research On Auto-Classification Topic-Specific Search Engine

Search Engine is a IR technology surged in 1990s.After over ten-year-long development, it has stepped into the live of people in all fields. It implemented many techniques and promotes them greatly, such as page rank, automatic classification, information extraction and query expanding. However, traditional search engines, namely general ones, take requirements of all users into consideration, whether they are searching for paper of computer related or news about basketball. At the same time , some techniques such as auto-classification used in general search engine are not ideal because of its fields are so broad.This thesis mainly analysis the principle and implementation of classification for search engine, especially the significance of it's implementation to topic-specific ones. We design an experiment for the three methods for classification, and it proves the topic data is easy to be classified.This thesis also focus on the design of topic-specific search engines in detail, and explain the implementation and significance of leading-word, pagerank, authority and hub pages, hyperlink anchor text analysis. This thesis also discuss the development of topic-search engine, specially the technology of the design of frame of the system, database and tables, robot and interface. At the end of this article the conclusion and improvement of this topic-search engine are given.