Research On Tool Recommendation System Based On Collaborative Filtering And Case Reasoning

Cutting tools play an important role in the manufacturing industry.However,with the diversified and personalized development of products,there are more and more types of cutting tools.The continuous increase of the types of cutting tools makes users have some problems in the process of purchase.How to help users recommend the required cutting tools from the system is the key problem to be solved.Adopt push + search two-way service.The push service adopts the user based collaborative filtering recommendation algorithm to recommend products according to the user’s historical information,but the recommendation accuracy is affected by the similarity measurement method,and the accuracy is not high.The search service adopts case-based reasoning technology to recommend the tools used to find similar cases.This paper designs the tool recommendation system from two-way service,which has good practical value in the field of tool recommendation.This paper takes the user’s acquisition of appropriate machining tools as the research object,and adopts collaborative filtering and case-based reasoning technology to accurately recommend tools to users.The similarity measurement method of traditional collaborative filtering algorithm and the case attribute weight optimization method of case-based reasoning are studied.Finally,a two-way service tool recommendation system including turning tool and milling tool is established through B / S architecture.The main research contents and conclusions of this paper are as follows:(1)The system actively pushes the service of cutting tools to users,which has the problem of low accuracy.The traditional collaborative filtering algorithm and similarity measurement method are fused to improve the recommendation accuracy.A tool recommendation algorithm for collaborative filtering based on Jaccard is proposed.Through the Jaccard similarity method and Pearson similarity method,the relationship between common scoring items is considered locally,and the Babbitt coefficient and KL divergence method consider the relationship between all scoring items of users globally,so the Jaccard similarity method,Pearson similarity method,Babbitt coefficient and KL divergence method are fused.Comparing the fused algorithm with the traditional algorithm,the results show that when the number of neighbors is 40,the MAE value tends to be flat,and compared with cosine similarity,modified cosine similarity and Pearson similarity,the MAE index is improved by 5.20%,2.20% and 2.76% respectively,and the RMSE index is improved by 3.75%,1.46%and 1.31% respectively,which improves the accuracy.Verify the effectiveness of the algorithm.(2)This paper studies the problem of providing users with accurate tools when they actively query Using case-based reasoning technology and grey correlation method as case retrieval method,the attribute index weight of grey correlation is optimized by BP neural network.An example is given to verify the effectiveness of the proposed method.Finally,the weight of the case is optimized based on the fact of the case,and the weight of the case is integrated into the weight of the case.Finally,the neural network is used to optimize the case,which is more in line with the requirements of the case.Through example verification,the results show that the most similar cases can be selected from the case base to verify the rationality of the proposed method.(3)The design and implementation of tool recommendation system based on two-way service are studied.In order to accurately recommend tools to users,the tool recommendation system framework is designed,the functional requirements of the system are analyzed,the corresponding database is established by mysql,and the tool recommendation system is developed and designed by Java,Vue and other development languages to realize the tool recommendation function.