The Research About The Intelligent Search Engine

With the rapid development of the network and computer science technology, many types of data (e.g. text, image, audio, video etc.) have appeared in the Internet. Besides, the amount of web information increase in exponential rate and the people's ability of store data are so limited. Although the forgetting rate of human can be reduced in some degree through professional trains, it can not reconcile this conflict. So, how to locate the required information rapidly and accurately in the Internet is a focused question which people are very interesting. The current technology of search engine can satisfy people's need to some extent, but it have many inherent defects to be attacked.The major works of this paper can be generalized with three aspects:1. Demonstrates the system structure and defects of the three kinds of search engine. Based on the analysis, the architecture of an intelligent search engine is proposed. This architecture absorbs the advantages existed in the independent or meta search engine. These improvements of the new system decrease the topic sensitivity and the scale of information interaction.2. Some methods are applied to increase the intelligence of the system. Firstly, through analyzing the static distribution and dynamic evolution of the user interest, an algorithm is proposed for constructing and adjusting user interest model. This model is used to obtain personal result list for one same query. Secondly, the paper specifies the affection of the tags to a query item and the traditional automatic abstracting technology is extended based on the Chinese syntax. Thirdly, in order to decrease the topic sensitivity, a dynamic strategy of the independent search engine set is proposed.3. Three most popular web page ranking algorithms are discussed. A new algorithm called A-PageRank is proposed in this paper. A-PageRank is one of the improved PageRank algorithms. It uses the set of anchor text as the substitute of the web page topic and the PageRank value of one source page is distributed proportionally to its link-out pages based on the topic similarity. At the same time, a series of experiments are carried out to prove the effectivity of the new algorithm.