Automatic Construction Method For Domain Concepts Based On Wikipedia Semantic Knowledge Base

With the development of network teaching mode,which is represented by the large open network course MOOC(massive open online courses),In order to meet the needs of learners fine-grained,high quality of learning resource retrieval requirements and use of information extraction method to construction a dynamic update for the machine algorithm using the domain concepts of semantic knowledge base,to support the MOOC system of learning resources is of great significance according to the semantic retrieval.There are some defects in existing semantic knowledge base including static and limited,which are difficult to meet the needs of large-scale network text information retrieval.Therefore,we studied the methods of automatically constructing and dynamic updating semantic knowledge base,which are as follows:(1)The method of combination about LDA topic model and TF-IDF algorithm is proposed better resolve these problems,which are extracting keywords not accurately representing the semantics of the concepts and not obviously be distinguished use information extraction method.We make full use of the advantages of LDA topic model to extract keywords semantic representation of comprehensive,TF-IDF algorithm to extract the keywords weight value distinguish obvious.Experiments show that this method can significantly improve effect than others.(2)To resolve problem that the semantic representation of concept knowledge base is not comprehensive,we use the keywords to represent the semantics of the concept itself,by hierarchy relationships of concepts in Wikipedia,link relationships between concepts and explanatory text of other concepts and link relationships between explanatory texts of different concepts to build concept semantic knowledge base.Therefore,link relationships between concepts and the semantic relationship of the concept itself are both used to construct concept semantic knowledge base,which are more comprehensive.(3)Graph's random walk algorithm is used to compute the semantic similarity between concepts in order to resolve the problem of inaccurate computation the semantic similarity between concepts.We make full use of the advantages of the random walk algorithm to achieve the stable value after several iterations of the probability value.The accuracy of this experiment can achieve more than 84%.The validity of using the random walk algorithm to compute the semantic similarity of the knowledge base is verified.(4)To resolve the problem that the concept node of semantic knowledge base scale is relatively small and need to be updated dynamically,we make full use of the advantages of Word2vec's Skip-gram model,which can skip some symbols.The dynamic update of knowledge base is achieved by acquiring a synonym to expand the number of nodes.