Research On The Construction Method Of Mongolian Knowledge Graph Based On WordNet

Knowledge graph is a kind of visual way to display all kinds of knowledge elements contained in the knowledge base.With the continuous innovation of science and technology in recent years,language information processing began to focus on the visual display method of knowledge graph.Natural language processing,especially Mongolian information processing,currently focuses on solving semantic problems.Therefore,studies of lexical semantic networks and knowledge graphs have become hot topics in the current study.All of the data in this article come from the WordNet underlying database,because almost all concept information is included in WordNet.At present,some ontology libraries,semantic dictionaries,and even Mongolian noun semantic webs are designed based on WordNet.This article uses WordNet as a data source to improve the utilization of resources.Compared with unstructured data,it reduces resource preprocessing,saves time,and speeds up construction.This article studies the key technologies in the construction method of knowledge maps based on WordNet.At the same time,this method is applied to the construction of knowledge graphs in the computer domain.The specific research contents are as follows:1.First analyze the parameters that affect the semantic similarity,after the comprehensive analysis of the advantages and disadvantages of each parameter,the information content as a parameter,putting forward a new IC value calculation model due to the existing algorithms.This article calculates the concept IC value based on WordNet's own structure while taking into account the depth of each concept of WordNet in the semantic tree and the number of conceptual child nodes.After experimental verification,the improved IC computing model is more in line with the characteristics of theWordNet semantic tree.2.After the semantic similarity parameter is selected,the existing algorithm is analyzed.Finally,combining the IC parameters,a new model of semantic similarity calculation is given.The model also takes into account the concept's semantic distance.Through the experimental test,it is found that the improved semantic similarity model calculates the semantic similarity value and the artificial score.The correlation coefficient is higher than other calculation methods,which shows that the algorithm model is superior to other calculation models.3.Extraction of concepts and relationships between concepts is the most critical step in constructing a knowledge graph.The extraction of the top level concepts in this paper is performed using an improved model of semantic similarity algorithm.Then the concept set and the relationship between concepts are obtained based on the WordNet underlying database.4.The construction of the knowledge graph of this paper is carried out according to the following process: based on the worker's division of the target domain into sub-areas that are not included in each other,and a core concept is set for each sub-area,and the semantic similarity algorithm is used to obtain each sub-domain top concept.According to the top level concept,an initial set of concepts for each sub-domain is obtained with WordNet's underlying database.Then use the semantic similarity algorithm to find the semantic similarity between the initial concept set of each sub-domain and the top-level concepts of the rest of the sub-domains,and obtain the final concept set of the target domain.The Mongolian WordNet was used to map the final concept set to the corresponding Mongolian vocabulary.Finally,using the graphic database theory,the constructed target domain knowledge map was visualized and displayed.5.The construction of Mongolian computer domain knowledge graph is a test of this article based on WordNet Mongolian domain knowledge graph construction method.It shows the specific construction process and will have certain reference value for the further improvement of Mongolian knowledgegraph.