Research on Knowledge Recommendation Technology Based on Domain Knowledge Graph: A Case Study in Aerospace Engine Domain

Abstract

In order to enhance knowledge reuse in product design and development, we propose a Domain-Specific Knowledge Graph-Based Recommendation Approach (DKGR) in conjunction with the Intelligent Knowledge Management System (IKMS) of an aerospace research institute in Beijing. The DKGR technique leverages the rich semantic relationships within the Domain Knowledge Graph, including product structures, task associations, and knowledge links and incorporates user logs into the DKG. This optimization helps address user matrix sparsity, resulting in improved accuracy and interpretability. Experimental analysis using real-world datasets demonstrates that the DKGR technique achieves an average F1 score of 0.515, compared to 0.343 for traditional recommendation algorithms. It indicates that the DKGR technique provides superior recommendation services in real-world scenarios.