| The project of Network Educational Resource Management System(NERMS) ,sponsored by Science Committee of Jilin Province and assigned to Knowledge Engineering Lab of the Institute of Computer Science and Technology in Jilin University, is a grand large-or-middle-scale project. The aim of the project is to organize and manage various kinds of educational resource effectively so that people can share and gain them efficiently. In order to satisfy the request that the system can provide intelligent answer services to its users, we make a research on Intelligent Answer System.Intelligent Answer System can receive or understand user's question by friendly alternation interface and question logic consequence parts, search knowledge base and information base to find question's answer or corresponding source bases computing methods and consequence. Info can more legibly present answer and information to user by answer's explaining part. System holds out some approach which solution problem. The knowledge base and information base of system have capability of auto studying and updating. System provides computing and statistical ability to corresponding data of answer's behavior; thereby optimize structure of system knowledge base and information base, and provides data's output.Intelligent Answer System simulates experts of some information domain for user's question to automatically provide exact answer. Some technique used by Intelligent Answer System generally includes Data Mining, Artificial Intelligence, Natural Language Processing and so on. The primary idea of this paper is searching Intelligent Answer System based on Natural Language Processing technique. Concretely, this system realizes natural language understanding by using LSA.In the course of Natural Language Processing, a powerful method and theory that processes a big number of glossary and information is needed, however LSA is a more appropriate means. We can accurately compute similarity between word and word, word and document, document and document. We open out their potential relation then implement Natural Language Processing and comprehending.Latent Semantic Analysis (LSA), a statistical technique that represents the content of a document as a vector in high dimensional semantic space based on a large text corpus, is used to predict how much readers will learn from texts based on the estimated conceptual match between their topic knowledge and the text information. During system's processing which answers to user, we analysis question in Semantic Space, then receive question's vector in Semantic Space and calculate the similarity between question's vector and document answer's vector. Finally, we find answer which matching user's question in answer storeroom. Constituting Semantic Space is the base of compute vector similarity, originally space is constituted basing on amount of Term-Document. Analyses documents and divides word from all documents, then gain all word included in documents and words' amount. Construct original semantic space according to words' amount and number of documents. Originally Semantic Space is decomposed into two spaces. One describes term semantic space and the other describes document semantic space. Whole computing in similarity's calculating is based on these two semantic spaces that decomposed by SVD method and theory.SVD is the general method for linear decomposition of a matrix into independent principal components, and is a form of Eigenvalue-Eigenvector analysis or principal components decomposition and, in a more general sense, of multi-dimensional scaling. SVD is a form of factor analysis, which constructs an abstract semantic space with n dimensions. Each originally word and document or each new document is described vector in space constructed by SVD decomposition. SVD decomposition computes singular value of matrix by algorithm of matrix, then receive two singular matrix by calculating which uses singular value, so originally matrix is decomposed into three other matrix...
|
PDF Full Text Request