Online Shopping Mall Recommended System Research And Application Based On Time Effects

M-commerce has been widely used in shopping malls, which provides people a lot of convenience in shopping experiences. However, the overload of shopping information affects people to select goods quickly and appropriately. The emergence of recommendation system can solve such problems in a certain extent. While users’interests change over time, which demands the dynamic recommendation of products. This paper presents a recommendation system which was on the basis of time effects, which generates recommended products dynamically according to the time change, in this way to support marketing precisely.On the basis of personalized recommendation technology, data mining, and some other related theories, this paper analyzed the influence of interest to customers caused by the time factor, focused on the study of collaborative filtering recommendation algorithm, which was based on time effects. Collaborative filtering recommendation algorithm recommends interested goods to customers which those who have the same interest. The time effect based filtering recommendation algorithm proposed in this paper first take the consideration of the diversity evaluation criteria among different customers, and the customer evaluations vary in different categories of commodities, the evaluation tendency has obtained via normalized treatment of goods evaluation, fixed customers evaluation matrix of goods, to this end, the Ebbinghaus forgetting curve inspirited that the customers’shopping behavior of forgetting their own history of shopping is in line with the nonlinear forgetting rules, This paper has introduced the concept of evaluation time window, that the influence of customers evaluation information remains steady in this time period, then by adding time window, this new parameter improves the traditional time weight function, to generate the new time weighted function, so that the customer evaluation of goods would be closer to the actual influence of current recommendations, By constructing the time weighted commodity evaluation matrix, the customer interest change with time migration has been solved. Then, the traditional clustering algorithm has been improved, by adding the time weight in calculating the similarity between customers, the current interest is quitely similar to the customer clustering, which makes the clustering more accurately,and the nearest neighbor search candidate set is constructed before collaborative filtering, which makes the algorithm more simple and easy to calculate, and can respond quickly, Through the customer based collaborative filtering algorithm, calculating the target customers’predictive evaluation of unevaluated goods according to the nearest neighbor customers’commercial evaluation, and during the prediction and evaluation stage, this paper also used new time weighted function to weight the evaluation of goods, then the forecast higher N evaluations of goods are recommended to the customer.The recommendation system developed in this paper has been used in a large shopping mall already. The shopping mall recommendation system centered on the time effect of collaborative filtering algorithm. By using the questionnaire form to the customer survey and evaluation of the recommendation system, which proves the effectiveness of recommendation system and the higher customer satisfaction.