Application Research Of Hybrid Recommendation Algorithm Based On K-Means In Big Data Environment

With the advent of the big data age,online shopping has gradually deepened into our clothing,food,housing,and travel.The offline tran-saction has gradually changed into online trading.We can solve the daily needs without going out of our homes.The development of the e-commerce platform has brought great benefits to our life.Each big e-commerce platform has a large number of browsing transaction data every day.These massive information conceals a lot of valuable data to users or businesses.These valuable data can be transformed to provide better services to users and businesses.A large number of data is from the log,database,or crawler of the platform,it is not processed.It is impossible to obtain valuable informa-tion from these data manually,and some scientific methods are needed to support it,the main recommendation algorithms are content based recomme-ndation algorithm,collaborative filtering recommendation algorithm,hybrid recommendation algorithm and so on.After analyzing and comparing the main methods of recommendation,it is found that because of the large amount of data and less evaluation data,it will lead to the sparse problem of the rating matrix,which will affect the accuracy of the recommendation algorithm.Therefore,the use of traditional recommendation algorithm can not bring better service to users,and can not create more value for businesses,and need more suitable algorithms to improve the quality of service.The main idea of the traditional collaborative filtering algorithm is to generate user-item rating matrix based on historical data to find users similar to the target users and to recommend them.The matrix can be viewed as a visual evaluation data from 1 to 5 in the field of movies,books and music.But in e-commerce environment,user rating matrix is not an explicit form of digital representation,but an implicit representa-tion based on user behavior.Aiming at the field of e-commerce,this paper is an improved collaborative filtering algorithm based on user network behavior profile,and uses improved algorithm to calculate the similarity between users.In this paper,an image model of user behavior is established by analy-zing the behavior of the user,the features of the goods and the interac-tive features of the user and goods.Based on the built user portrait model,an improved collaborative filtering algorithm is proposed.This algorithm combines the user's online shopping behavior and the time based user's online shopping behavior frequency to calculate the similarity between users.Then it integrates with K-means clustering algorithm,and comp-letes user recommendation based on fusion algorithm.Finally,by calcula-ting the accuracy of the algorithm,the recall rate and the F1 value,the experimental results show that the accuracy of the algorithm is greatly improved compared to the classical collaborative filtering algorithm,which proves the effectiveness of the proposed algorithm.