摘要
基于购物活动表层挖掘的推荐系统的时效性和信息持续性较差。为解决相关问题,提出了基于客户心理挖掘和预测的推荐系统,给出了该系统的解决方案、结构模型以及处理流程。该系统采用多维向量空间存储心理特征数据,并使用贝叶斯算法对客户与商品进行聚类;采用基于功率谱估计的心理特征预测算法生成推荐商品选择。实验结果表明,该系统具有较好的信息持续性,并能够较准确的进行推荐活动。
Due to traditional shopping record mining, the recommender systems bearded bad real-time and information persistent. In order to deal with these problems, a novel system is presented based on client psychology mining and predicting. And its＇ system schemes, structure models and processing flows are given as following. Then it utilized the multi-dimension vector space to store psychology character data and the Bayesian algorithm to cluster clients and commodities. Topic power spectrum estimation detection method is used to predict clients＇ current emotion and generate recommended commodity lists. Simulation resuits show that the system has better information persistent and accurate recommendation than the traditional does.
出处
《计算机工程与设计》
CSCD
北大核心
2012年第11期4347-4351,共5页
Computer Engineering and Design
基金
教育部人文社会科学研究基金项目(10YJCZH169)
四川省金融智能与金融工程重点实验室基金项目(FIFE2010-P05)
西南财经大学校管课题基金项目(2010XG068)
华侨大学科研基金项目(07HSK02)
关键词
推荐系统
客户心理
挖掘
预测
特征
recommender system
client psychology
mining
predict
character
