The Research On Recommendation System For E-commerce Based On Combinatorial Algorithm

The rise and development of internet,and then,leads to the rapid development of e-commerce.E-commerce,which brings a lot of data to us.However,faced with these data,users can not quickly access to the effective information,which induces to the problem of information overloading.Search engines have made some successes in solving this problem,such as Google of America and Baidu of China are both search engines.When the users enter the keywords in the search engine,the search engine will find the information which the users may need in the database.However,with so many data in the internet,it is not every user can clearly know what they need.By contrast,the recommendation system is a better way to solve the problem of information overloading,which is also more intelligent and active.The recommendation system does not require the users to enter the keywords,which can carry on the fast search among the magnanimous data information.On the other hand,the recommendation system can actively present the users with the information which the users may be interested in,and,according to the different interests of different users it can provide the users with personalized recommended services.Therefore,in many filds,they will use the recommendation system,especially in e-commerce sites,which have a growing prospective of development.There are many recommendation algorithms,but among which the most popular and the highest utilization rate are three recommendation algorithms: content based recommendation algorithm,collaborative filtering based recommendation algorithm and association rule based recommendation algorithm and so on.Among them,the most widely used recommendationalgorithm is collaborative filtering based recommendation algorithm.However,with the rapid development of internet,the number of users and products are also increasing rapidly.The recommendation algorithm based on collaborative filtering also has exposes more and more problems to solve,among which the most concerned are data-sparsity and cold-start.How to solve these problems has always been a problem for researchers.In order to make recommendation algorithm based on collaborative filtering more effective,this paper proposes a combinatorial algorithm to compensate for these defects.On the other hand,when there are too much data accumulated in the e-commerce recommendation system,the data processing by the single computer is affected to a certain extent,which will also affect the accuracy and efficiency of the recommendation.Therefore,using the distributed processing to handle these data,using the hadoop technology to achieve the calculation of the large amounts of data in the recommendation system,improving the efficiency of the calculation,then recommending the more accurate goods to the users,making users increasingly rely on e-commerce recommendation system,and making e-commerce wensite acquire some economic benefits.This paper mainly researches the following aspects :1)Several common recommendation algorithms are introduced in detail,and the ideas and implementation steps of each algorithm are described.At the same time,the shortcomings of various recommendation algorithms are discussed,especially the problem of collaborative filtering recommendation system.2)Using combinatorial algorithm to solve the problems of data sparsity and cold start.According to different conditions,different recommendation algorithm are choosed.3)Using mapreduce framework of hadoop to achieve the distributed computing of massive data of the recommendation system,and to make the recommendation system have a better performance.