Research On The Application Of Information Collection And Classification Oriented To Space Knowledge Management

With the booming development of domestic space industry and in the age of internet, big data and knowledge economy, space industry institutions which are knowledge-intensive began to introduce theories and technologies for knowledge management to cope with mass information resources. According to the circumstance and requirements, they start to build appropriate knowledge management system for effective mining, organization, management, utilization and storage of core knowledge resource in space domain. Acquirement of knowledge is an indispensable and basic procedure in knowledge management system. The design and application for acquirement of knowledge need many key technologies relating nature language processing and data mining, such as information collection, text classification, information extraction, knowledge map, semantic network and so on. Recently, the technology for knowledge acquirement and processing has rapid development both in theory and application filed. This thesis will conduct the research of two kinds of key technology for knowledge management system in space industry, which are information collection and text classification.The space industry knowledge is redundant and complex. It comes from both the massive documents of related companies and scientific institutes and the enormous internet information resource. According to these features of space information, the key technologies of effective knowledge management are effective and accurate information collection and efficient information management. Thus we can further conduct the tasks of information extraction and knowledge mining. Efficient space information collection will meet scientists and engineers’ needs for mass, professional, new, complete and accurate information. It will avoid useless work and cut down the costs for information acquirement. Automatic space information classification can help manage the complex and disordered information efficiently and accurately. It will help build space knowledge data base rapidly and optimize the information organization and search result of the information search system. Thus, scientists and engineers can conduct data mining task further. Therefore, the research of information collection technology and text automatic classification technology for space knowledge management in this thesis have significant value in practical application.The main research content in this thesis is as follows.First, this thesis introduces the background and significance of two kinds of key technology for space knowledge management, which are information collection technology and text classification technology. It also analysis the development situation of knowledge management and its application in space domain as well as the development situation of information collection and text classification technology and their application in space domain.Second, this thesis studies the space information collection method based on theme crawler. Other study include designing general framework of space information collection system based on theme crawler, accomplishing the programing task, accomplishment of two kinds theme certification model which are method based on theme vector space models and method based on Support Vector Machine classifier, constrasting the experimental results with the method based on key words matching.Third, this thesis studies space text multiple classification algorithm based on SVM. Other study include designing multiple classification procedure framework for variety kinds of space information, improving the method for features weighting by increasing the weight of distinguished words such as space terms according to the features in space domain to optimizing classification result, accomplishing the relate programing task, contrasting the experimental results with the text classification method based on naive Bayesian model and the method based on K-Nearest Neighbor.The experiment results demonstrate that the information collection and text classification method for space knowledge management in this thesis meet the expected performance. The space information collection method based on theme crawler in this thesis is effective and efficient. It can satisfy the needs for information acquirement in space domain. The improved text automatic multiple classification algorithm based on SVM for space knowledge management in this thesis is effective and efficient. Compared with other classification methods, it has more outstanding performance and could satisfy the needs for effective and accurate automatic text classification in space domain.