Research On Improvement Of The Performance Of Search Engine

This paper, first, introduces some new technology and research focus of the current search engines, and then begin to talk about the two research topic:the queries' understanding and classification; the sorting method of search result based on ontology. For the first topic: the queries' understanding and classification, on the foundation of previous researches, we have presented two query class functions respectively based on the search engine's click-through data and anchor link data. We also founded two mathematical models: Bias model and Normalized Entropy model. They have implemented Large-scale query classification. And then, this paper used one of the Machine Learning's methods: Maximum Entropy to classify the queries automatically. We have a lot of discussion about how to collect the effective classification features. The experiments results show that the classification is essentially to meet the needs of real application.Next part of the paper is about the research of the sorting method of search result based on ontology. Ontology has the best ability of knowledge description, and also has the features of other description models. This paper first outlines the concept and basic principles of Ontology, second, proposed a detail Ontology-based document ranking function, and then given application results of these methods in practical. The results show that this function we proposed has an indeed value in the specific area search.