Learning Algorithm Of Search Engine Based On The Small-World Network

The study of network has long history. At very fast speed development of Internet, we more and more depend on it to get the information. It is a crucial problem that how to get information effectively.The "small-world" effect was highlighted in work by Milgram 1967.He described an experiment -pass letters from acquaintance to acquaintance, from which he deduced that many pairs of apparently distant people are actually connected by a very short chain of intermediate acquaintances and the length of this chain is about six. In more recent work, D.J.W. and S.H.S have used the phrase "small-world network" to mean a network which has short paths. In their opinion, Internet, World Wide Web and many network in our real world are all "small-world network".Since the Web is small-world network, if two pages belong to a connected component of the Web, some short path exists between them. How can we find the short path with only local information? How to help the users get what they really need from a large number of Web pages? Content and link information is used by virtually all search engines to crawl, index, retrieve, and rank Web pages. The correlations between similarity measures based on these cues and on semantic associations between pages is therefore crucial in determining the performance of any search tool.There are three parts: The first part of this thesis studies several representatives of search engine, arithmetic of search engine and common evaluation methods. The second part of this thesis studies the properties of the small-world network, the model of small-world network, navigation arithmetic. And from three similar methods: content similarity, link similarity, semantic similarity analyzes the properties of the small-world network. The second part of this thesis gives the arithmetic of search engine, and improves precision with the " the small-world network " property.