Research On E-commerce Personalized Recommendation Based On MapReduce

ABSTRACT:With the advent of the era of big data, the user’s personal information was in different storage device in a variety of ways, integrated all the user information and through certain mining technology to potential demand of users. Current e-commerce development fast, mobile e-commerce will be more dominant in the future, how to excavate potential demand of the user’s personal rapidly, the users may be of interest to push the products to become the era of big data, electronic business enterprise needs to solve the problem. The current e-commerce personalized recommendation precision needs to be improved, personalized recommendation means a single, not after a deep analysis of the data, but is based on the user’s browsing information and purchase information to users recommend related products. The recommended way of efficiency is not high; the personalized recommendation of the era of large data frame should be from source analysis and mining.In view of the above problem, this article from the origin of large data sets, according to the different means of user information sources, collection of all the personality information, and then use the era of big data mining technology, the associated information, and stored in the database of the enterprise, for enterprises to carry out related products recommended. To build a personalized e-commerce recommendation system based on graphs, and gives the core modules in the mining system, based on the thought of graphs, puts forward the idea of dividing the data set, referred to as APD algorithm, effectively avoid the scanning of the entire database, improve the efficiency of the mining, and through the global optimal solutions and local optimal solution is obtained based on the theory of graphs the correctness of the segmentation algorithm of APD. Based on the APD algorithms, this paper puts forward the idea of a gradually eliminate, by setting support degree and confidence, put forward step by step in each individual data blocks is not in conformity with the set of frequent item sets, the level of until the optimal results, this paper gives the algorithm of thought process, the process and the algorithm of demonstration and prove that can draw the algorithm saves time and space resources, is with high efficiency, can satisfy the personalized recommendation to the requirement of time and space resources under the big data.In future electronic commerce recommendation, especially, when the mobile e-commerce, under the big data mining technology will be able to greatly improve the accuracy of electronic commerce recommendation, will obtain good effect, the solution actual problem, to help marketers and find the right marketing mix strategy, to reduce the cost. Improve marketing success rate and profit; dig out the potential relevance and laws, for the retail business system provides the theory basis for scientific decision-making.