基于Hadoop的电子商务推荐系统设计与实现

发布时间：2018-04-27 02:26

本文选题：大数据 + 电子商务　；参考：《西安工业大学》2017年硕士论文

【摘要】：山西大同某游戏公司开发了具有当地特色的多款棋牌类游戏,同时以游戏为依托多元化经营,提供了免费送货服务当地百姓的日用品网络商城电子商务平台,但近年来经营出现了用户数量停滞、效益逐渐下滑的现象,通过长期考察和了解,发现用户的需求和喜好已经发生了变化,如何动态的准确把握当地百姓的消费需求已经成为公司亟需解决的头等大事,研究适合当地百姓的大数据电子商务推荐系统成为解决此问题的最佳选择。为满足用户需求和提供优质服务,论文设计了基于Hadoop的电子商务推荐系统,推荐系统根据用户点击的路径分析出用户的需求,从过去对交易后数据分析来发现用户需求过渡到现在对交易前的过程数据分析,从而挖掘出用户的个性化需求,并及时为每个用户精确推荐满足需求的物品,从而提高销售业绩,为用户带来收益。针对数据海量性,推荐要求精确性、及时性等这些问题,本论文设计了基于Hadoop平台的电子商务推荐系统整体框架。对大数据平台、推荐算法、电子商务平台进行了需求分析,采用在Linux系统上搭建Hadoop大数据平台;利用分布式文件存储系统HDFS和MySQL混合数据库构架结构化和非结构化的数据存储;采用计算模型MapReduce构架起电子商务构架推荐算法的引擎,并以此实现基于用户协同过滤算法、基于物品协同过滤算法和混合推荐算法,同时设计和实现了电子商务推荐系统功能,并将推荐结果通过用Tomcat的Web服务器以电子商务网站的方式呈现给用户。当前系统已经上线试运行,未来将在后续数据采集上更丰富些、个性化商品推荐上更精确、更及时。
[Abstract]:A certain game company in Datong, Shanxi, has developed a number of chess and card games with local characteristics. At the same time, relying on the diversified operation of the game, it has provided free delivery services to the local people on the e-commerce platform of the commodity network mall. However, in recent years, the number of users has stagnated and the benefits have gradually declined. Through long-term investigation and understanding, it is found that the needs and preferences of users have changed. How to dynamically and accurately grasp the consumption demand of local people has become the top priority of the company. The study of big data E-commerce recommendation system suitable for local people has become the best choice to solve this problem. In order to meet the needs of users and provide high quality services, this paper designs an E-commerce recommendation system based on Hadoop. The recommendation system analyzes the needs of users according to the path clicked by users. From the past analysis of post-transaction data to the transition of user requirements to pre-transaction process data analysis, the personalized needs of users are mined, and timely and accurate items are recommended for each user to meet their needs. In order to improve sales performance, for users to bring revenue. Aiming at the problems of magnanimity of data, accuracy and timeliness of recommendation, the whole framework of E-commerce recommendation system based on Hadoop platform is designed in this paper. The requirements of big data platform, recommendation algorithm and e-commerce platform are analyzed, and then the Hadoop big data platform is built on the Linux system, and the structured and unstructured data storage is constructed by using the distributed file storage system (HDFS) and MySQL mixed database. The engine of E-commerce framework recommendation algorithm is constructed by MapReduce, and the function of E-commerce recommendation system is designed and implemented based on user collaborative filtering algorithm, article collaborative filtering algorithm and mixed recommendation algorithm. The recommended results are presented to the user by using Tomcat's Web server as an e-commerce website. At present, the system has been put into online trial operation, in the future will be more rich in the follow-up data collection, personalized product recommendation more accurate, more timely.
【学位授予单位】：西安工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3

【参考文献】