王小豪　孙彦武　胡浩明　王蕊蕊　乐光学

摘要：商品推荐的发展不仅是大数据和云计算技术发展的成果体现，也是电子商务行业的发展趋向，商品推荐是电商行业的关键技术，是提升电商行业经济效益的有力途径之一。通过在电商行业应用商品推荐技术，能够实现快速搜索和推送目标商品，优化用户购物体验，推动新时代电商行业的进一步发展。因此，该文通过对现有商品推荐算法的分析和研究，尝试提出了基于商铺和商品信誉度的推荐算法。通过问卷调查确定影响商铺和商品信誉度的关键指标，运用Entropy算法和欧几里德距离分别计算商品信誉度和相似度，最后根据信誉度和相似度完成商品推荐。通过建模和仿真分析，该算法在一定程度上提高了商品推荐的可信度和精确度。

关键词：商品推荐;商铺/商品信誉度;Entropy;欧几里德距离

中图分类号：TP391    文献标识码：A

文章编号：1009-3044（2019）13-0294-03

Abstract： The development of commodity recommendation is not only the result of the development of big data and cloud computing technology， but also the development trend of e-commerce industry. Commodity recommendation is the key technology of e-commerce industry and one of the powerful ways to improve the economic benefits of e-commerce industry. By applying commodity recommendation technology in the e-commerce industry， it is possible to quickly search and push target commodity， optimize the user's shopping experience， and promote the further development of the e-commerce industry in the new era. Therefore， this paper attempts to propose a recommendation algorithm based on the reputation of shops and commodities through the analysis and research of existing commodity recommendation algorithms. Through the questionnaire survey to determine the key indicators affecting the shop and commodity credibility， using Entropy algorithm and Euclidean distance to calculate the commodity credibility and similarity， and finally complete the commodity recommendation based on credibility and similarity. Through modeling and simulation analysis， the algorithm improves the credibility and accuracy of commodity recommendation to a certain extent.

Key words： commodity recommendation; shop/commodity reputation; Entropy; ED

1 引言

近年来，随着计算机、互联网及社会化媒体技术的广泛應用，电子商务得到了快速发展，数以亿计的消费者在电商的服务下无时都在产生着海量的交互信息。对于信息过载的电商系统，如何在琳琅满目的商品中给消费者推荐目标商品，已成为当前研究的重点课题[1]。

目前，较为成熟的电商推荐系统包括基于协同过滤的推荐系统[2]，基于内容的推荐系统[3]，基于网络结构的推荐系统和基于混合模式的推荐系统[4-5]，非主流、新兴的推荐系统（比如基于统计、效用的推荐系统、神经网络、贝叶斯网络等[6]）。……