Research And Implementation Of Automotive Parts Supply Recommendation System Based On Knowledge Graph

Due to the large number and diverse types of automotive components,it has become an important issue for the production of automobiles to enable units to obtain the necessary components in a timely manner.The implementation of a knowledge graph based automotive parts supply recommendation system can help demand units timely obtain automotive parts information and supplier information,and on the other hand,enable suppliers to improve sales efficiency,thus achieving a win-win situation for both the demand and supply sides of automotive parts.Recommendation algorithms or models are widely used in e-commerce,entertainment,and other fields,but existing recommendation algorithms or models still have some shortcomings.Because the traditional collaborative filtering algorithm and its improved algorithm have data sparsity and cold start problems,it is unable to accurately describe user characteristics.Due to the current recommendation algorithms or models not fully mining the rich association relationships between entities,it is difficult to mine the potential needs of users.The recommendation performance of existing recommendation algorithms or models on the automotive parts dataset in this thesis needs to be improved.In response to the above shortcomings,this thesis studies and implements an automotive parts supply recommendation system based on knowledge graph and graph convolutional neural network algorithm.This thesis mainly includes the following aspects:1.Build the automotive parts supply recommendation model based on knowledge graph.The model includes five modules: human-computer interaction module,data preprocessing module,knowledge graph building module,recommendation module,and testing module.2.Build a knowledge graph.To construct a knowledge graph,this thesis first needs to preprocess demand unit data,automotive component data,supplier data,and order data.Calculate the similarity of demand unit data;Data fusion of automotive component data and supplier data to eliminate redundancy and duplicate information;Construct a scoring matrix for order data.The preprocessed data is identified as entities,so that the entities are connected with each other.The entities are connected through relationships to form a "entity relationship entity" triplet,and the triplet is stored through the Neo4 j diagram database for easy query and use.Based on this,a network knowledge structure is formed to obtain a knowledge graph of automotive components and a knowledge graph of demand units.3.Provide a knowledge graph based recommendation method for automotive component supply.This thesis studies the recommendation algorithm based on knowledge graph and graph convolutional neural network(KGCN＿UI)to achieve recommendation of automotive components.Firstly,query the knowledge graph,find a set of nodes,and collect contextual data related to each node.This data is obtained from the knowledge graph,including the hierarchical structure type,roles,attribute values,and inferred neighbors of each contextual neighbor node.Then,this information is processed into a set of arrays,which are aggregated and combined one or more times as vectors for constructing nodes.Then calculate the predicted value by combining the combined vector with the user vector and item vector.Finally,by calculating losses and continuously updating iterations,the algorithm can converge and generate recommendation results.This algorithm uses a knowledge graph as the data carrier,utilizes the advantages of graph convolutional neural network feature extraction,and optimizes it by combining item based and user based recommendations,resulting in improved recommendation performance.In order to verify the effectiveness of the algorithm,this thesis conducts comparative experiments using three classic algorithms: SVD,Ripple Net,and KGCN.The experiments show that the recommendation algorithm based on knowledge graph and graph convolutional neural network performs well in recommendation performance.4.Implement and test a knowledge graph based automotive component supply recommendation system.By providing a knowledge graph based automotive parts supply recommendation model and algorithm,this thesis uses My SQL to store data,uses Django to build a website framework and write code to implement the system,and conducts running tests on the system to ensure the normal flow of automotive parts supply recommendation system business.Through algorithm comparison experiments,it has been shown that the recommendation algorithm based on knowledge graph and graph convolutional neural network proposed in this thesis can alleviate the problems of cold start and data sparsity,fully explore the relationships between entities,and achieve a certain improvement in recommendation performance.Through system testing,ensure the normal operation of the knowledge graph based automotive component supply recommendation system.The research results of this thesis can improve the sales efficiency of automotive parts,achieve collaborative sharing of network collaborative manufacturing resources,and have certain reference value for supply recommendations in other industries.