Research And Implementation Of Clothing Personalized Customization System Based On Collaborative Filtering

With the rapid development of e-commerce,the dependence of clothing products on ecommerce platforms is becoming more and more obvious.However,faced with many resources,it is difficult for consumers to pick out the right clothes,and most of the clothing orders are still in a state of "people cater to clothes." In order to meet the user's pursuit of individuality and selfawareness,online clothing customization system came into being.At present,the development of domestic online clothing customization system is still in its infancy,the customization mode is relatively simple,the customization efficiency needs to be improved,and the personalized service function of the customized system cannot be fully reflected.With the expansion of the system scale,the number of commodities is more and more abundant,and users will also face the selection problem brought by information overload in the customization process.Therefore,providing personalized recommendation service in the customization process is the focus of this paper.At present,there are few researches on clothing recommendation in the e-commerce environment.The clothing product recommendation faces many problems,the experiment is complicated,the recommendation accuracy is difficult to improve,etc.In this paper,the most widely used collaborative filtering algorithm is studied,and collaborative filtering for clothing products is proposed.The main work is as follows:(1)For the traditional collaborative filtering algorithm,the data is sparse,the precision is not enough,and the similarity calculation method can not reflect the user's true similarity in some cases.An improved trust-based collaborative filtering algorithm ATUCF is proposed.The concept of trust is introduced,and the trust factor is calculated according to the number of common scores between users and the number of respective scoring items to ensure an asymmetric trust relationship between users.The calculation method of improving the trust degree further highlights the trust relationship between users,and finally trust factor is fused to calculates the nearest neighbor set accurately.The comparison algorithm verifies the accuracy and effectiveness of ATUCF.(2)Aiming at the shortage of clothing recommendation in the e-commerce environment,the coverage is low,and it is difficult to mine the potential interest of users.A clothing recommendation algorithm PCUCF that integrates user preference trust is proposed.Through the literature research,clothing data collection and expert knowledge summary,the paper establishes the clothing attribute library.The Chinese natural language processing technology is used to extract the attributes of the clothing goods,and the noise information in the non-attribute library is removed.A user-to-clothing preference model is established based on the user-attribute scoring matrix,and the difference between the positive preference and the negative preference is calculated as the preference trust.Finally,the weight parameter is combined with the score trust degree and the preference trust degree,and the nearest neighbor set of the user is selected and a clothing recommendation list is generated.The experimental results show that not only the accuracy of the algorithm is improved,but also the potential interest in mining users.(3)In view of the current simplification of clothing customization mode and low efficiency,research and implement a clothing personalized customization system incorporating recommendation functions.In-depth specific needs analysis of the user's personalized clothing customization,determine the overall design of the system and various functional modules,apply the clothing collaborative filtering recommendation algorithm proposed in this paper to the system,complete the system development and function verification.