基于BP神经网络的情境化信息推荐服务研究  

Research on Contextual Information Recommendation Service Based on BP Neural Network

作　　者：周朴雄[1] 张兵荣[1] 赵龙文[1] 

机构地区：[1]华南理工大学经济与贸易学院,广东广州510006

出　　处：《情报科学》2016年第3期71-75,共5页Information Science

基　　金：国家级社会科学基金项目项目(14BTQ009)

摘　　要：随着移动互联网的发展,个性化信息推荐研究逐渐开始将用户情境作为信息推荐中重要的影响因素,但目前的研究多数是停留在如何将用户在不同情境下的资源偏好融入到推荐系统中,缺少对不同情境中各个要素对信息推荐的影响研究。由此,本文建立了结合情境的用户偏好模型,提出运用BP神经网络方法来预测不同情境下的用户对资源类别的偏好,同时结合项目协同过滤算法来实现精准的个性化信息推荐。最后将试验结果与传统的项目协同过滤算法对比,推荐准确率有了提高。With the development of mobile internet, the researches on the personalized information recommendation gradu- ally began to take user context as the important factors influencing the information recommendation, but most of the current researches are how integates users＂ different situations of resources preference into the recommendation system, the re- searches of the influence of various elements in different situation on information recommendation are lacking. Thus, This paper established a user preference model which combined with the situation, proposed to predict different situations of us- er preferences on the categories of resources by the method of BP neural network, and combined the collaborative filtering algorithm to achieve personalized information accurate recommendation. With the comparison of experimental results and collaborative filtering algorithm, the recommendation effectiveness is improved.

关 键 词：BP神经网络 项目协同过滤 移动网络环境 情境化信息推荐 

分 类 号：G252[文化科学—图书馆学]