Research On Key Issues In Subject-oriented Knowledge Element Indexing

In the knowledge economy era, knowledge has attracted the attention of many scholars. The scientific management and effective application of knowledge can achieve value-added knowledge, which comes to be one of the most important targets about knowledge management in digital era. In order to provide better knowledge services, it is required to research knowledge organization and knowledge management from knowledge element level. Based on the knowledge organization and management relevant theories, this dissertation builds up a data model for the knowledge element, gives a solution for the extraction of guide information about knowledge element and a method of knowledge element indexing, and provides a realization for this solution and implementation, such as design and programming project for this solution. The main contents of this dissertation include:Firstly, a subject-oriented knowledge element data model is constructed.After the explanation of the knowledge element definition, we choose text knowledge element as the research object, and build up a subject-oriented knowledge element data model according to the object-oriented design methods, the model includes nine foundational elements of knowledge element. A XML document is created to represent the data model information to describe the text knowledge element of the subject.Secondly, the method of guide information extraction from document based on Neighbour Words Co-occurrence is provided.The guide information means subject word. From the analysis for the partly chapter documents of journal, we obtain the properties of the subject words which are indexed manually. Based on the related linguistic properties of texts such as frequency, property, context characteristics and location, we provide the method of subject word extraction from document based on Neighbour Words Co-occurrence and give some basic definitions. The experiment results show that our method need not subject dictionary, and the result is better than the traditional TF/IDF.Thirdly, the method of knowledge element indexing based on regulation is provided.Choosing four typical elements in knowledge element data model including knowledge element name, description, property and source, we analyze the general construction characteristics by manual and find the sentence regulation. According to the manual analysis result, we build the extraction regulation of knowledge element, and provide the method of knowledge element indexing based on regulation. The experiment result shows that our method can extract the main knowledge element description and improve greatly the efficiency of the knowledge element indexing. Furthermore, we take a research about the knowledge element relationship.Finally, a subject-oriented knowledge element indexing system is designed and implemented. The dissertation has detailed the whole system design, flow chart, database, realization, testing and its evaluation, making it work properly with the main functions of indexing system of knowledge elements. This function contains the text preprocessing, extraction of subject words, extraction of description sentences of knowledge element, presentation and the retrieval of the knowledge element. This prototype system integrates the above three aspects and inversely proves the possibility and efficiency of our research work. We describe the application research of knowledge element-based clustering retrieval system for educational text resources.The main results of this dissertation include:(1) A subject-oriented knowledge element data model is constructed. This data model is different from traditional information organization model which is based on resource, index and meta-data catalog. This data model gives the knowledge element structure, which is the foundation to build a "theme, knowledge element and topic map" knowledge organization model.(2) A method of subject word extraction based on Neighbour Words Co-occurrence is provided. The method is provided according to the linguistic properties of texts, the subject context and the characteristics of subject word mining, which realizes the subject word automatic extraction.(3) A method of knowledge element indexing based on extract regulation is provided. Different from the information and resource indexing method in large granularity, this method combines the semantic content information and subject word relationship, and discovers the knowledge element from document by rules, and realizes the knowledge element indexing in small granularity and semantic.The research results of this dissertation concentrate the foundation for the research of knowledge element-based knowledge mining, knowledge integration and knowledge enrichment key technologies research.