Hypergraph Ontology Sparse Vector Representation and Its Application to Ontology Learning

Abstract

Data representation is the first step in computer science modeling, and the performance of the algorithm directly affects the efficiency of the algorithm. For structural data, graphs and hypergraph representations are often ideal choices, which can effectively represent the connections between data. The edge weight function can also be used to express the strength of the connection, and the adjacency matrix and the Laplacian matrix describe a kind of motion, and the eigenvalues of the matrix reveal the motion trajectory of the vector. In this paper, hypergraph framework is used to characterize the internal structure of the ontology sparse vector, and use the hypergraph theory to define a new hypergraph Laplacian matrix, and thus obtain the control items on the sparsity of the ontology sparse vector learning model. Then, a new algorithm is proposed. Finally, two specific implement tests are used to verify the effectiveness of our new proposed hypergraph based ontology learning algorithm.