电商平台林产品个性化推荐算法研究

发布时间：2018-04-13 19:43

本文选题：林产品推荐 + 协同过滤　；参考：《东北林业大学》2016年硕士论文

【摘要】：林产品以其具有的天然、环保、绿色的优势,成为了健康产品的主流选择之一。电子商务平台的不断发展带动了林业产品推广和销售。但是随着用户和产品规模的不断增多,出现了严重的“信息负载”问题,因此个性化推荐服务应运而生。个性化推荐服务能够快速主动挖掘潜在的购买用户,帮助用户快速找到可能感兴趣或喜欢的商品,不但可以增加网络流量、提升营业收入,同时还能够加强用户对于网站的忠诚度以及用户体验。电子商务平台对于产品的推荐大多是基于协同过滤推荐算法,该算法是迄今为止应用最成功的个性化推荐算法,被广泛的应用到很多领域中。但是随着互联网的快速普及,使得电商平台用户、产品规模的不断扩大,协同过滤算法遇到严重的数据稀疏性问题,导致推荐的精度和可扩展性都在急剧下降。文中对协同过滤算法进行深入学习和研究过后,提出了一种基于Weighted SlopeOne(简称WSO)的K-means个性化林产品推荐算法,该算法首先将WSO算法进行产品打分的思想应用于高维稀疏用户-产品评分矩阵的填充上,然后使用改进的K-means算法对用户进行聚类生成用户类簇,最后在每个类簇内为目标用户实现推荐服务。文中以MovieLens数据集为数据源进行对比试验,经仿真表明,文中的算法能够有效地提升推荐的精度和可扩展性。以Apache Mahout为实验平台,将文中提出的基于WSO的K-means个性化林产品推荐算法应用于京东商城的林产品购买评分中,实验结果表明,文中提出算法的precision、recall、MAE指标反应良好,适宜将文中提出的算法在林产品贸易销售平台进行大范围推广,以提升林产品销量和用户忠诚度。
[Abstract]:Forest products with its natural, environmental protection, green advantages, has become one of the mainstream choice of health products.The continuous development of e-commerce platform has led to the promotion and sale of forestry products.However, with the increasing of users and products, there is a serious problem of "information load", so personalized recommendation service emerges as the times require.Personalized recommendation services can quickly and actively mine potential buyers, help users quickly find products that may be of interest or interest, and not only increase network traffic, but also increase revenue.At the same time can also enhance the user's loyalty to the site and user experience.The product recommendation of e-commerce platform is mostly based on collaborative filtering recommendation algorithm, which is the most successful personalized recommendation algorithm so far, and has been widely used in many fields.However, with the rapid popularization of the Internet, the users of e-commerce platform and the product scale are expanding, the collaborative filtering algorithm meets with serious data sparsity problem, resulting in a sharp decline in the accuracy and scalability of recommendations.After in-depth study and research on collaborative filtering algorithm, a K-means personalized forest product recommendation algorithm based on Weighted Slope one (short for WSO) is proposed.This algorithm first applies the idea of WSO algorithm to the filling of high-dimensional sparse user-product scoring matrix, and then uses the improved K-means algorithm to cluster users to generate user clusters.Finally, the recommendation service is implemented for the target users in each cluster.In this paper, the MovieLens data set is used as the data source. The simulation results show that the proposed algorithm can effectively improve the accuracy and scalability of the recommendation.On the basis of Apache Mahout, the K-means personalized forest product recommendation algorithm based on WSO is applied to the forest product purchase score of JingDong Mall. The experimental results show that the proposed algorithm has a good response to the index of precisioning all forest products.It is suitable to popularize the proposed algorithm in forest products trading and marketing platform in order to promote forest product sales and customer loyalty.
【学位授予单位】：东北林业大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.3

【相似文献】