Research On Thesis Subject Indexing Management System Of Reading Oriented

Through analyzing the disadvantages of traditional knowledge management system and users' requirement, this thesis research put forward the viewpoint that is the method of knowledge management must go deeply into the inside of text and index kernel of knowledge paragraph. And the system must achieve the hierarchical mapping of knowledge—concept node—subject indexing—text, and based on this structure, the system can increase the precision of knowledge management enormously.Concerning data structure of the system:A document extraction structure was adopted to record reading's essentials.And the document subject structure was adopted to describe thesis,which comprises main subject and subtopics.The field conceptual network was adopted to represent basic field knowledge of thesis. Based on the above data structures, the system can record the hierachical structural data.As regard to subject indexing method:It carried forward a subject indexing method which is based on the concept node position relationship to calculate subject radius, this method takes the paragraph which was marked by reader as indexing object, and based on the indexing result of every paragraphs, it calculates the indexing result of whole document. Therefore ,the system can achieve its goal which is that indexing method must index kernel of knowledge paragraph. The method take the steps of calculating subject radius, removing the interferential nodes, adjusting node's weight, calculating the document subject structure as the result. The precision of indexing method was proved by the experiments. In the end, the system adopted a process of result modification and conceptual expansion, so users can reconstruct the conceptual network with users cognize.As regard to the knowledge searches method: This thesis has analyzed the expectations of thesis searches, and designed a thesis knowledge searches method which is based on document subject structure indexing result.We used the experiments to compare two searches result, it came to the conclusion of the method can searches knowledge of kernel of paragraphAs to the personalization of system: According to the readers' habits and the characteristics of the system, this thesis put forward a method to calculate personalization. By using the similar modification factor, it will be able to calculate the users' profiles. Meanwhile the method uses Web Services as the approach of information acquisition. The feasibility of this personalized method was proved by the experiments conducted in this study.The prototypal system was constructed the four hierachical mapping, and achieved the goal of indexing the kernel of paragraphs.