| During the information construction of universities, different departments choose respective environment of software and hardware based on their own information requirement and special applications. Therefore, various data are organized and managed according to departments and functions. That leads to the solitary of data resources and services of colleges and universities.Facing to these information isolated islands, the sharing method of media resources based on description information integration is studied in this paper, and the sharing model of media resources is built. The model draws lessons from Web services and P2P's partial complementary characteristics. System uses coordinator's centralized services in the Web services and marginalization distribution of resources in the P2P model to realize files exchange in campus network. System uses XML to unifiedly describe media resources, and puts the all of distribution information into the directory server to manage together. So the directory server can receive the query of users. But the media resources are put in the users' terminals and managed by users, through that way servers' load are reduced and flexibility ratio of resources management is enhanced.Sharing information of media resources are stored in server by means of XML. For improving retrieval efficiency of XML, building index is an effective method. For meeting the needs of XML retrieval with complex structure and multiple quantity, system builds an efficient "document-keyword-node" two-level index structure based on inverted index. The structure includes abundant structure and content information of XML without significantly increasing spaces of index and reduces searching range of index, so retrieval efficiency of XML is improved. A calculation method of correlation degree based on TF-IDF is put forward in this paper. According to the calculation method of result correlation degree in the existent information retrieval theory, this method considers the influence of the appearance position of XML keywords and tree structure of XML documents. By that, the retrieval efficiency of XML is improved.According to the study upon, the system of media resources sharing and retrieval in colleges and universities is designed and realized. The system realizes the functions of the media resources issue, media resources retrieval and media resources maintenance. The system was tested and verified, and then the system finished the functions and achieved the design goal. |