Recommendation Algorithm of Product Configuration based on Knowledge Graph Optimization

To satisfy the demand for automated product configuration in manufacturing enterprises, a product configuration recommendation algorithm based on knowledge graph optimization is proposed. Firstly, the historical configuration information of the same product, material and component is mined through FP-Tree, and the inventory data of the same configuration relationship is compressed into a single normalized weight data. Secondly, combined with the weight data, the knowledge graph is constructed by using the a priori relationships of the same product, material and component stored in the structured database. On this basis, the recall of associated materials and components is completed by using the graph query of product entity, and then the FM algorithm is used to achieve detailed sorting recommendations based on the normalized weight data. Validated using production data, the accuracy rate of product configuration recommendation reaches 74.54%.