Visual retrieval method for multivariate graph database based on attribute enhanced representation learning

• Main Authors: SU WEIHUA, CHOU DAVID, LIU YUHUA, SUN LING, WANG YIGANG, WANG HAOXUAN
• Format: Patent
• Language: Chinese; English
• Published: 14.01.2022
• Subjects: CALCULATING; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention discloses a multivariate graph database visual retrieval method based on attribute enhanced representation learning. The method comprises the following steps of: firstly, extracting structure and attribute characteristics of a multivariate graph by utilizing a graph representation learning model and a mathematical statistical method, establishing a graph representation learning model based on attribute enhancement by utilizing canonical correlation analysis in combination with a characteristic extraction result of the multivariate graph, and fusing a structure vector and an attribute vector into a comprehensive embedding space; and then projecting the high-dimensional structure-attribute fusion vector into a two-dimensional space and clustering to construct a distance-based graph retrieval model. Visual evaluation is conducted on retrieval results from the structural similarity and the attribute similarity through a node link graph and a parallel coordinate view, interaction is designed to help