基于大数据的Web个性化推荐系统设计  

Design of Web personalized recommendation system based on big data

作　　者：张婷婷 ZHANG Tingting(School of Information Engineering,Shandong Management University,Jinan 250357,China)

机构地区：[1]山东管理学院信息工程学院,山东济南250357

出　　处：《现代电子技术》2018年第16期155-158,共4页Modern Electronics Technique

基　　金：国家自然科学基金青年项目(71301086);山东省社科规划专项基金(17CQXJ11);山东省高等学校科技计划资助项目(J16LN70)~~

摘　　要：为了解决基于数据挖掘技术的Web个性化推荐系统对Web的推荐结果准确率低,反应时间长的问题,设计基于大数据的Web个性化推荐系统。塑造系统组成框架图,设计系统的总体功能包括源数据采集、数据预处理、用户兴趣分析与实现、个性化推荐以及推荐引擎。源数据采集利用Sqoop工具将数据库中的数据转移到HDFS中以便H-ICRS算法进行数据提取,并获得推荐的历史数据,实现作为系统上层数据支持的功能。针对分析用户长远和当前的Web兴趣度,分别采用语义分析模型和分片聚类的方法,分析用户Web使用兴趣。塑造单个推荐引擎的推荐引擎架构,得到最终的Web个性化推荐列表。实验结果表明,所设计系统的Web个性化推荐结果准确率高,系统的抗压能力强。A Web personalized recommendation system based on big data is designed to solve the problems existing in the Web personalized recommendation system based on data mining technology for its low accuracy rate of Web recommendation results and long reaction time.The composition framework of the system is built.The system′s overall functions including source data acquisition,data preprocessing,user interest analysis and implementation,personalized recommendation,and recommendation engine are designed.During source data acquisition,the Sqoop tool is used to transfer data in the database to the HDFS,so as to extract data by using the H-ICRS algorithm,obtain the recommended historical data,and realize the upper layer data support function of the system.By analyzing users′long-term and current Web interest degree,the semantic analysis model and fragmentation clustering method are adopted respectively to analyze users′Web interest.The recommendation engine architecture is constructed for a single recommendation engine to obtain the final Web personalized recommendation list.The experimental results show that the designed system has high accuracy rate of Web personalized recommendation results and strong antipressure capability.

关 键 词：大数据 HADOOP WEB个性化推荐 系统设计 Sqoop H-ICRS算法 

分 类 号：TN919-34[电子电信—通信与信息系统]