Research And Application On Intelligent Search Engine Focus On Petroleum Information Service

With the rapid development of Internet and WWW, the resource in Internet become rich more and more,computer network has become the largest knowledge database in the world. This has provided a perfect platform for the share of information resources, but also created difficulties for inquiry of user's original willing. The general search engines are very excellent at their search function, but they are unfitbale in a certain professional field, research-result often include useless-information. "Information overload" caused a great of inconvenience for people accessing to useful information. Search engine, as the information seraching tools on Internet, has been made continues progress, and become more and more professional, automatic and intelligent.The subject faced petroleum-information-field. In the analysis of the features and principles of professional search engine, the system combined with the feature of oil, researched these technologies and functions, such as information-filter, text-classification and information-related calculation. In the aspect of information collection, this paper designed a kind of vector space mode for calculating information-related degree. While capturing information on web, it will filter low related-degree information automatically and reduce the redundant data. In the aspect of word segment, in order to resolve the problem that in the course of word-marching computer run slowly and consume lot of memory, this paper proposed a fast segmentation method based on key-tree. This way improved the efficiency of the word segment in a great extent. This paper also create a method for splitting word without dictionary, by this way, high frequency words will be made available to the administrator and the dictionary will be expanded automaticly in order to ensure that petroleum information include completely. In the aspect of text-classification, with the shortage of traditional classification model, this paper proposed a FVSM based on text feature vector, this model improved the accuracy of classification. In the aspect of user-information-research, this paper used XML to build an initial knowledge base of petroleum information. The system analysed users'search terms and searched information from the concept net in order to improve the precision of information and the rate of recall.