The Research On An Improved Algorithm For Specialized Search Results Rank

With the development of Web technology,Internet already become the important approach to acquire information and communicate.Although search engine have already turned into the main tools for searching informations in various fields of our life, it is still hard to find the specialized content which needed by the users,due to the existed many deficiencies in separating different search topics.In order to solve this problem,specialized search engine is developed with the increasing requirement.It pay attention to the in-depth portrait service,take up with the full information of a certain special field,the rest of information unrelated with the specialize topic will not be embodied.Nowadays,the research on most specialized search engines is still on testing period.The domestic research on specialized topic searching method are seldom processed the development is slow and the research group is small.Many users are not satisfied with the research result,the rank of search results have an important effect on search capability and satisfication.Therefore research on the rank method of specialized search engine is useful and has a great meaning in academic field.Specialized search engine is integrated topic with universal search engine.At first, some ranking algorithms of universal search engines such as the word frequency and location algorithm,Direct Hit algorithm,PageRank algorithm and Hits algorithm are analyzed systematically in this paper.Secondly,the improved rank algorithms domestically and abroad research are analyzed and compared.Meanwhile,the improved methods based on existing PageRank technologies is also discussed in this paper.One of Specialized search engine's kernel technologies relevance algorithm is introduced. This paper discuss the relevance calculating model:Boolean Model,Vector Model, Probability Model and their characteristics.Then the topic influencing rank algorithm is also analyzed.At last,some factors such as the special positions,state and frequency where the querying words stand,feedback from users,links,text length,e-value selection and iterative formula are discussed.Then the topic relativity factor,improved from document classify,kinds of dimensionalities which given different weight in topic,the variety of different relevance between document and topic,fellowship links and so on are processed in this paper.Experimental results show that this improved ranking algorithm work outperforms the original one in query precision.Finally,kinds of factors are appraised particularity in the algorithm.Search Engine is still a developing research topic,and many related technologies are not mature yet.We will improve the research result in the future.