| Primary education knowledge graph is a project under construction.This paper is one of the small branches of the work-through the establishment of connectivity,trying to explore and show the relationship between concepts,even between concepts of different disciplines,in order to facilitate the further development of subsequent teaching applications.The specific research contents are as follows:When creating the primary education knowledge graph,we need to segment the corpus in the database.Because dictionaries can easily and efficiently divide the corpus,this paper uses the dictionaries in hanlp word breaker to segment the words,but there are some words not included in the primary education of dictionaries,that is,the unregistered words.In order to improve the success rate of knowledge graph connectivity construction,it is necessary to add unlisted words in the field of primary education to the dictionary.In this paper,the n-gram method is used to segment the original data,the word frequency is counted,the threshold value is set to select the seed words,and the left and right entropy and mutual information method are used to filter out the unregistered words,so as to further improve the dictionary.In the process of building the connectivity of primary education knowledge graph,this paper proposes an A* algorithm based on mixed semantics,and compares the Bi-Breadth First Search algorithm.In the heuristic function of the algorithm,a cosine similarity algorithm based on part of speech weight is proposed to solve the problem of cosine similarity.In this algorithm,part of speech weight is added to improve the relationship between entities,and relevant parameters are introduced to adjust the error.Finally,different conditions are set to verify the proposed algorithm.In the case of adding login words and not adding login words,the analysis is carried out from three aspects: time,degree and connectivity.The results show that in the experiment of building the connectivity of basic education knowledge map,the cosine similarity method of part of speech weight used in the A* algorithm based on mixed semantics is far better than the Bi-Breadth First Search algorithm in terms of time and connectivity.Experiments show that under the existing conditions,the A* algorithm based on mixed semantics can complete the task of connectivity construction. |
