基于日志挖掘的网站分类目录用户心智模型研究

发布时间：2018-07-17 15:25

【摘要】：“以用户为中心”是电子商务网站取得成功的前提,而电子商务网站中有80%的用户倾向于使用分类目录,分类目录已经成为用户进行信息获取的重要途径之一。此外考虑到用户需求、兴趣等个性化的差异,大量的电子商务网站都引入了“个性化推荐”的功能,其实质为将最符合用户知识结构的分类目录推荐给用户。可见分类目录是电子商务网站的重要组成部分,因此从目录体系的设计到基于目录的个性化推荐都应做到“以用户为中心”,这就需要解决以下三个问题：如何获取用户关于网站分类目录的认知、如何在网站分类目录设计中体现用户认知、如何结合分类目录基于用户认知实施个性化推荐。本研究针对这三个关键问题,探索新的用户研究方法,直接利用客观体现用户心智模型的网站日志数据,从中提取用户关于网站分类目录的认知,从用户心智模型概念相似性出发,基于聚类分析与路径搜索法,分析用户内心期望的分类目录体系与网站分类目录体系间的关系,进而为构建更符合用户心智模型的网站分类目录体系提供决策支持；从用户心智模型概念空间性出发,借鉴协同过滤和基于内容推荐算法的思想,应用聚类分析与多维尺度法,基于网络日志分类用户,挖掘三级目录之间的相关性以给用户个性化推荐相关产品主题(即最相关的三级目录),尝试给用户带来良好的个性化推荐使用体验,从而为电子商务网站留住更多的用户。
[Abstract]:"User centered" is the prerequisite for the success of e-commerce websites, and 80% of users in e-commerce websites tend to use classified catalogues. Classified catalogues have become one of the most important ways for users to obtain information. In addition, a large number of e-commerce websites have been introduced into account of user needs, interest and other personalized differences. The function of "personalized recommendation" is actually to recommend the classified catalogues that are most consistent with the user's knowledge structure. It is obvious that the classified catalogue is an important part of the e-commerce website. Therefore, from the design of the directory system to the personalized recommendation based on the directory, it should be "centered on the user", which needs to solve the following three questions. Question: how to obtain the user's cognition about the classified catalogue of the website, how to embody the user's cognition in the design of the website cataloging, and how to implement the personalized recommendation based on the user cognition in the classified catalogue.
This study aims at these three key problems, exploring new user research methods, using the web log data that objectively embodies the mental model of the user, extracting the users' cognition on the classified catalogue of the web site, starting with the similarity of the mental model of the user, based on the clustering analysis and the path search method, to analyze the classification of the users' inner expectations. The relationship between the catalog system and the web classified catalog system, and then to provide decision support for building a web catalog system that is more consistent with the user mental model. Based on the concept of user mental model, the ideas of collaborative filtering and content based recommendation algorithm are used for reference, and the clustering analysis and multidimensional scale method are applied to the network log classification. Class users, mining the correlation between the three levels of directory to personalize the users to recommend related product topics (the most relevant three levels of directory), and try to give users a good personalized recommendation and use experience, so as to retain more users for e-commerce sites.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092

【参考文献】