Scheme Design And Implementation Of Dress Collocation

Today, e-commerce platform includes all aspects of people's basic necessities of life. The platform also strongly requirement Recommendation System can provide efficient and accurate recommend. At present, dress collocation recommend sets off a craze in the field of Recommendation System. This paper studies the dress collocation recommendation, and provides some feasible recommendations. The main contents of this paper includes:(1) A method is proposed to provide popular collocation recommendation, based on an improved FP-Growth Algorithm. It uses the law that the difference of clothing is great in different seasons, and uses association rule algorithm in a continuous period history data, and achieves a higher accuracy recommendation based single rule. At the same time, the improved FP-Growth Algorithm saves the time and space;(2) A method is proposed to provide personalized recommendation, based on Resource Allocation of bipartite graph. It uses the degree of influence among users, which probability propagation mechanism reflects, to provide personalized recommendation;(3) A method based on the user's historical data to provide the user's preference collocation is proposed. This method explores the user's preferences by observing the user's historical behavior, to provide user's preferences collocation style;(4) A method based on the above two recommendations is proposed, to apply both lavish and personalized recommendation. The method is a combination of popular collocation and personalized collocation, and the final recommendation is obtained by sorting the preference style of individual users.In this paper, three recommended techniques are used to achieve dress collocation recommendation, and it analyzes the advantages and disadvantages of the various recommended techniques. The experiment proved that people dress collocation is influenced by user's personal preferences and clothing popular degree of influence; most users have their own clothing preference category, and sorted hybrid recommendation according to the users preference category, can improve the accuracy of recommendation; user collocation behavior will change along with the time. Finally, based on the experimental results, this paper summarizes some improvement methods,which may improve the recommendation accuracy.