Research On Recommendation Algorithm Based On Commodities' Attribute Weights

With the rise of online shopping,a novel and effective recommendation system become more and more important.A good recommendation system will help find goods suit for customers' hobby in the shortest time,thus to build a good customer experience.However,traditional recommendation algorithm,often fail to provide recommendation effectively for the reason that customers' inner motivation behind the purchases is usually ignored.Actually,customers' purchase is driven by their own preferences.How to find the customer preference and apply it to the model building of recommendation systems,will make a difference to the recommendation algorithm's effectiveness.Optimization model of the clustering based on attribute weights as reference,recommendation algorithm based on commodity's attribute weights,the indicator of customer preference,is designed.Taking A Ecommerce Platform as an example,we test the efficiency of new algorithm and get the following conclusion.The model is meaningful: First,it has expanded the scope of recommendation services from a single individual to the group under a particular rule,and thus help alleviate the cold start problem that people who lack of shopping online record can't be offered with the recommendation service appropriately;Second,a new way based on goods' attribute weights to optimize recommendation systems is provided,it has proved that the introduction of commodity's attribute weights can help improve the accuracy of recommendation.In terms of A Ecommerce Platform,the new algorithm can significantly improve the accuracy and the efficiency of recommendation systems,having a better performance compared with the former algorithm.