Research On Construction Of Course Knowledge Graph And Search Technology

The development of semantic Web relevant theory as well as technology and deepening of research on knowledge graph and semantic search laid solid foundation for the ultimate realization of Web of data. The essence of knowledge graph is a kind of semantic network, which described the entities and relationships between entities in reality. By means of knowledge graph and search system, it is able to understand user query better and provide users with the contents that are closer to query terms. Meanwhile, users can clarify the relationship between object and entity, and improve the efficiency of obtaining knowledge. The construction of knowledge graph mainly involves in the work of two levels:one is to organize at schema level and the construction of schema is the extraction of knowledge; follow the given model not only is beneficial to the standardization of knowledge but also is conducive to the subsequent process such as knowledge search. Another is to enrich and extend entity into ontology at data level; extract structural information from structure-free or semi-structural web, relate it to entity and establish the relationship between entities. The semantic search based on knowledge graph is an intellectualized search; since the extra relevant structural information is offered except for existing search results, it can satisfy more precise information search demands; moreover, the form of search results developed into the search answers based on entity and relation limited to the links matching search keywords.There is a large number of course data in the Web related to education field; due to the limitation of traditional electronic learning platform, it cannot realize the correlation between different course data at data level. The traditional data organization method not only hindered the share of information and data at semantic level, but also made it difficult to the interoperability of heterogeneous data. In view of this, this article proposed the organization method of course knowledge graph by combining with the semantic technology recommended by Web international organization and the experience of electronic study system developed by this research group; the correlation of entity and relation in course resource by using the associated data at data level changed the traditional organization and release method of curse data. The process and handling of entity related concept evolved it into knowledge ontology so as to lay foundation for the intellectualized search service.Based on semantic Web, knowledge graph, and theory as well as technology of semantic search, it researched the organization of course data knowledge graph and entity search method. The research is mainly embodied in the following aspects:(1) Aimed at the technical characteristics and resource organization scope related to data level, it analyzed the flow, strategic method and technology of associated data organization. According to the features of data in education field, it raised a new organization method of associated course data that composed of data conversion, data association, storage and index, and data application; through this method, the traditional teaching resource can be used to construct the associated data that is in accordance with international standard, which is beneficial to knowledge integration and reuse.(2) Comprehensively researched on the construction flow of knowledge graph and elaborated the method of ontology learning at schema level and entity enrichment at data level. Based on analyzing the advantages and disadvantages of former entity enrichment method and combining with the features of course data, a new method is proposed; it processed entity similarity by adopting the attribute value of adjacent points in the resource description frame diagram; by comparing with outgoing link, calculating triad distinguishing degree and introducing the optimization based on distance, it confirmed the enrichnment of entities through context information.(3) As for the semantic search based on graph, it researched on the features and methods of ontology search, graph search and entity search. Under the background of knowledge graph, it raised the expansion search method by using ontology and similarity of entity; combining with personalized and contextual information to provide users with more accurate results. While designing new semantic expansion algorithms, ease data sparseness problem in the process of the search, improved search accuracy.(4) The experiment of course knowledge graph and search mainly selected the table resource with high quality semantic relation as object. The specific process is to map the columns of table as appropriate category firstly, relate the unit value to entity or character constant secondly, then conform the relationship between columns and generate appropriate RDF triples, and complete the inference of semantic knowledge as well as generate course knowledge graph in graphical model through message passing algorithm and by using LOD resource as background knowledge. The transfer approach in this experiment avoided the large amount of calculation of joint probability distribution and promoted the understanding of semantic connotation in the table. By expressing the schema and instance in the table according to select available URI could complete the complex task of generating course knowledge graph. Search results show that table conversion methods conducive to the integration of information resources, while improving the user experience of satisfaction.To sum up, the deep research of the organization and search of course knowledge graph made by this article involved in the process of entity at data level in knowledge graph, ontology learning of schema level, entity semantic search and other specific contents. According to the characteristics of course data, it proposed new conversion as well as organization method; the experiment based on industrial authority data set manifested that the research in this article is effective with certain innovation.