个性化推荐算法分析及在电子商务领域应用-计算机应用技术专业论文.docx

哈尔滨工业大学工学硕士学位论文 哈尔滨工业大学工学硕士学位论文 II II Abstract As the rising of online shopping, people have received more and more unprecedented convenience. However, with the number and size of e-commerce sites continues to expand, customers need more time to find what goods they want to buy. On the one hand, people face a lot of information helplessly, and are often subjected to information overload; on the other hand, e-commerce websites may lost contact with the users, and cannot establish a solid relationship with them. Solving these problems focuses on transforming passive acceptance of visitors into active speculation of the visitors’ information needs for the requests of e-commerce websites, to realize active information services to visitors of e-commerce system. Thus, personalized recommendation systems come into being. In the increasingly competitive environment, personalized recommendation systems can effectively prevent losing of site users and improve the sales of the e-commerce websites. The traditional recommendation algorithms can be divided into the following categories: content-based recommendation algorithms, collaborative filtering recommendation algorithms, knowledge-based recommendation algorithm and combining recommendation algorithms. However, there are several issues existed in these recommendation algorithms, such as cold-start, sparse, scalability, and so on. In this paper, we used mobile e-commerce systems as our experimental platform, employed the frontier technology of HTML5 to carry out the cross-platform developments, and proposed an dynamic threshold personalized algorithm aiming at the shortcomings of the recommendation algorithms existed, which optimized the recommendation systems on the platform e-commerce, and added new factors according to the characteristics of the mobile platforms, to enhance the recommendation efficiency. The work of this paper is listed as follows: Studied and compared the popular recommendation algorithms, and analyzed th