【作者】 王平；

【导师】 卜佳俊； 王灿；

【作者基本信息】 浙江大学 ， 计算机应用技术， 2007， 硕士

【摘要】 随着互联网的迅速发展，互联网用户和数据都在迅猛增长，从而导致了用户需求和海量信息之间的不对称问题。如何为用户提供准确而有价值的信息便成为一项具有重要应用价值的课题，而电子商务则为该课题的研究提供了一个良好的平台。为用户提供个性化的信息服务需要解决两方面技术问题，即如何获取用户偏好和如何获取分类信息。现代电子商务所提供的丰富而干净的数据使得用户偏好的挖掘成为了可能；同时，主题搜索技术的发展，也使主题相关的专业信息获取日趋精确与成熟。将数据挖掘和主题搜索技术相结合，将能有效的实现为用户提供个性化信息服务的目标。本文所描述的情报推荐系统有效的结合了用户偏好挖掘和主题搜索这两种关键技术。作为电子商务的一种辅助服务，情报推荐系统能够为用户提供有效的参考信息，能对用户的商务行为进行有目的的引导。本文的情报推荐系统主要包括情报分类、用户偏好挖掘和主题搜索三大模块。系统通过基于向量空间的情报分类模型对情报分类和用户信息需求进行了有效表达，通过对一定时期的历史数据进行数据挖掘获取用户偏好，并对应用户偏好从主题搜索结果中选取情报，最后将情报加以编辑后向用户进行推荐。同时，论文的实验数据将从技术细节上展示对系统性能和效率的优化改进，这些技术细节包括：改进的Apriori算法的运算效率、基于向量的数据挖掘模型的匹配准确度和基于向量的主题分类器的表现等。更多还原

【Abstract】 The asymmetry between the Web user requirement and massive Web information is becoming an escalating problem with the rapid increase of the Web pages and users. Consequently, technologies to provide customized information search are booming like flowers after a spring rain. Among them is the customized information push based on user preference mining in e-commerce. All transactions of e-commerce are stored on the servers, which builds a solid ground for data mining. While we can mine with a certain precision users’ preference, personalized recommendation services can be easily implemented. On the other hand, focused search technology is now capable of clustering topic-related information. Combining the two technologies to provide personalized information recommendation is the main theme of this paper.The Information Recommendation System (IRS) presented in this paper is based on user preference mining and focused information search and serves as a complementary service of e-commerce to provide guidance information to Web users. The IRS is comprised of three modules, namely information classified module, user preference mining module and focused search module. The IRS do a good integration with two technologies through good design, and two technology both use vector space model technology to be achieved. The system design of based on vector derived from the vector-based information retrieval technology, the technology in the information retrieval field to be changed and improved, thus becoming the technology adapting to the system. First, the system mines a certain period of historical data to obtained user preference. Then, corresponding user preference the system selects the information from the focused search’s result. Finally, the system recommends the editorial information to the user. Meanwhile, the experimental data showed by the paper display that the system has good performance: improved Apriori algorithm greatly improved the operation efficiency, Vector-based data mining model with a higher matching accuracy, Vector-based focused search classifier with better performance.更多还原