基于用户行为的用户全局兴趣模型构建研究

发布时间：2018-04-13 15:30

本文选题：个性化 + 用户行为　；参考：《武汉工程大学》2015年硕士论文

【摘要】：随着计算机技术的飞速发展和互联网的普及,人类所处的信息时代正发生日新月异的变化,其中主要原因是因为人类自身行为的复杂化及对于信息服务的更高的要求与期待。人类复杂的网络行为产生了更多的不规则的数据,同时却希望得到更加个性化的信息服务,从技术以及商业的角度而言,这是一种矛盾。全面理解用户的行为,判断用户的兴趣度与兴趣点,构建精准的用户全局兴趣模型(UGIM)是解决这一矛盾的最有效方法。无论是学术界还是商业运用,UGIM的研究从来都是一个热点;只有构建了科学合理的UGIM,准确体现和表达出用户的兴趣点,才能以此为基础为用户提供有针对性的个性化的服务,同时为运营商家的决策提供更加科学的依据,把更多的资源提供给更有价值的客户,从而解决个性化的需求与复杂的数据、高昂的成本之间的矛盾,达到信息的提供者与使用者双赢的目的。本文基于上述背景,以大学研究生教育创新基金项目课题为支撑,将理论演绎与实际运用相结合,以在线学习平台为原型系统,以机器学习和分类模型为研究范式,对以下内展开研究:1.总结概述了网络环境下用户行为的相关理论。综述了传统的用户行为定义及分类,归纳总结了用户行为的收集与分析方法。指出了传统数据收集与整理的不足,并着重介绍了最新的、基于网络开放平台的数据的收集与使用。2.对“用户全局兴趣模型(UGIM)”相关理论进行系统的描述与分析,综述了其内涵、分类、定义、表示方法等相关研究的成果,并重点介绍了基于本体论的表示方法及聚类算法,奠定了本文的原型系统模型构建的理论依据。3.实例演示用户行为数据的收集与整理,基于本体论表示方法的UGIM构建过程,结合实际的业务思考,分别从用户的兴趣度、用户兴趣点以及用户的潜在消费能力与意愿三个方面构建了三类实用的UGIM,每一类UGIM都以本体论的方式加以表示,其计算过程遵循k-nn聚类思想。实验结果表明,UGIM的查全率与准确率都达到了理想的效果,并且此思路流程较为清晰,考虑的用户行为指标较为全面,切实以实际运用为指导思想进行研究的,故研究UGIM的运用也很有必要。4.UGIM的运用研究,主要对UGIM在电子商务系统中的运用进行了研究,一方面通过其运用使系统能较为准确的预测到用户的所感兴趣的内容及程度,可以进行智能推荐;另一方面也可以从信息提供的角度把有效的资源针对性的提供给更有价值的客户。
[Abstract]:With the popularization of computer technology and the rapid development of Internet, the information era is changed, the main reason is because of the complexity of human behavior and information services for higher requirements and expectations. The complexity of the human network behavior produced more irregular data, but at the same time hope to get more personalized information services, from technical and business perspective, this is a contradiction. A comprehensive understanding of user behavior, interest and interest the user's judgment, the construction of user interest model global precision (UGIM) is the most effective way to solve this problem. Both academic and commercial use UGIM, the research has always been a hot spot; only to set up a scientific and reasonable UGIM, accurately reflect and express the interests of users, can we provide users with targeted Personalized service, and provide more scientific basis for the operation of the business decision, to put more resources to provide more value to the customers, so as to solve the needs of individual and complex data, the contradiction between the high cost of the information providers and users, achieve double win goal. Based on the above background, the University Graduate Education Innovation Fund projects to support the theory deduction combined with practical application, the online learning platform for the prototype system, based on machine learning and classification model as the research paradigm, research on the following: 1. summarize the related theories of user behavior in the network environment. The paper reviews the definition and classification of traditional user behavior. Summarized the methods for collecting and analyzing user behavior. It is pointed out that the lack of data collection and sorting, and introduces the latest, open platform based on Network The data were collected with the use of.2. in "global user model (UGIM) theory for the description and analysis system, summarizes the connotation, classification, definition, representation of the achievements of related research, and introduces the ontology representation method and clustering algorithm based on established theoretical basis and arrangement of.3. example user behavior data to construct a prototype system model of this collection, ontology representation method based on the UGIM construction process, combined with the actual business thinking, respectively, from the user's interest degree, the three aspects of potential consumption ability and willingness to user interest and user constructs three kinds of practical UGIM, each type of UGIM by the way of ontology is represented, and the calculation process follows the k-NN clustering algorithm. The experimental results show that the recall and accuracy of UGIM have achieved the desired results, and the thinking process is more clear Considering the user behavior index, comprehensive, practical for the study of the guiding ideology in the actual application, the use of the study UGIM have been using.4.UGIM, mainly for the use of UGIM in electronic commerce system are studied, one can accurately predict the content and degree of interest to users by using the system, can carry out the intelligent recommendation; on the other hand can also provide information from the perspective of the effective resources to offer more value to customers.

【学位授予单位】：武汉工程大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：F713.36

【相似文献】