Intelligent Generation and Reasoning Method for Aviation Assembly Knowledge Graph Based on Joint Knowledge Representation Learning

Abstract

The field of aviation assembly is characterized by high system complexity and nontrivial operation. Currently, the knowledge assistance in this field is mainly based on assembly technical manuals, and the document format is mostly unstructured, lacking structured descriptive information. Therefore, the application of knowledge-assisted technology faces problems such as insufficient accuracy of structured knowledge construction and interpretability of knowledge-assisted reasoning. To address the above problems, this paper proposes a knowledge graph intelligent generation and reasoning framework based on joint knowledge representation learning. Traditional keyword-based semantic retrieval knowledge assistance is insufficient for global knowledge and context reasoning. In the proposed framework, the prior domain knowledge is embedded in a large language model through knowledge representation learning for intelligent generation of structured knowledge graphs. Then, the generated knowledge graph will be vectorized in a low-dimensional space in the form of subgraph sets by knowledge representation learning, facilitating knowledge reasoning in the given domain. Verification Experiments of aviation assembly knowledge graph construction and auxiliary decisionmaking are conducted. The results show that the accuracy of proposed knowledge graph generation reaches 91.33%, the accuracy of knowledge reasoning is 96.55%. This method provides an efficient technical solution for knowledge management and application in the field of aviation assembly.

Keywords: Knowledge graph, large language mode, knowledge representation learning, aviation assembly, intelligent manufacturing