Research Of Clothing Recommendation System Combining Explicit Data And Implicit Data

With the advent of the "Internet +" era,the e-commerce platform represented by Taobao has rapidly emerged.In recent years,the proportion of clothing online shopping sales accounted for the first place in the online shopping sales of various commodities.The clothing online shopping platform generated a large number of users and clothing data in the process of sales.The recommendation algorithm can extract key information from the massive data and recommend clothing for the user.However,the traditional clothing recommendation algorithm is gradually unable to meet the needs of people's growing personalized clothing wear.The clothing recommendation system can divide the data into explicit data and implicit data according to the data type,and adopts a combination of two data types.The system can effectively solve the problem of clothing information overload,improve the recommendation accuracy and data utilization,and satisfy the requirements.The user has personalized wear requirements.The clothing recommendation algorithm combining explicit data with implicit data convertd the user's explicit feedback and implicit feedback into explicit data and implicit data,and combined user-based collaborative filtering recommendation algorithm to recommend clothing for users.Among them,the data preprocessing part,the explicit data adopted the user's preferred color and style data,the implicit data used the user to purchase records,browses the recorded data.Multiple data type matrix calculation part,extract data in the database to construct user feature vector and scoring matrix.the matrix processing part,due to the large amount of data of users and clothing,the dimension of the user-apparel matrix is too high,and dimensionality reduction is required to get the feature matrix.the algorithm used the optimized Pearson similarity calculation method to calculate the user similarity,and obtained the initial recommendation result.The recommended list processing part performed data processing on the initial recommendation result,and the processed data is sorted according to the TOP-N sorting algorithm.The top 10 garments werethe final recommendation.In the process of studying explicit data,an color clustering algorithm was proposed.Based on the popular color of the clothing collected by experts,the clustering center was calculated to form a user color library for users.Feedback on their favorite colors.This paper designed a clothing recommendation prototype system combining explicit data and implicit data,and proved the advantages of the two data type recommendation algorithms through experiments.The prototype system used the springcloud microservice architecture to develop the prototype system,and used Python to write experimental algorithms and performed algorithm verification.In the experiment of the clothing recommendation algorithm combining explicit data and implicit data,the experimental results showed that the recommendation algorithm based on explicit data(user rating)F1 value was 0.284,based on implicit data(purchase record)recommendation algorithm.The F1 value was 0.265,and the F1 value of the recommended algorithm combined with the two data proposed in this paper was 0.321.