PCA-GAT: A Knowledge Graph Approach for Machining Process Plan Recommendation

A new method called PCA-GAT (Constraint-Aware Graph Attention) has been proposed to address the challenge of integrating heterogeneous industrial knowledge for machining process plan recommendation. The approach formulates the problem as a knowledge graph enhanced collaborative filtering task, using Bayesian Personalized Ranking as the learning objective and Recall@K and NDCG@K for evaluation. The knowledge graph provides semantic structure when collaborative signals are sparse. Four domain constraints—material compatibility, precision requirements, feature applicability, and operation sequencing—are introduced as attention mechanisms. The work is published on arXiv with ID 2607.24213.