动态推荐技术的研究及在个性化电子警务中的应用

发布时间：2019-05-21 20:35

【摘要】：现代通信技术、Internet技术和多媒体技术的快速发展,在使用户能够快捷获取形式多样信息资源的同时,也不可避免的带来了“信息过载”( Information Overload)的问题。Internet媒体的使用或传递者,通过他们对资源的偏好、社会关系、任务协作等关联起来,形成了一种新型的推荐关系网络。为这种推荐关系网络提供感兴趣的信息,是减少信息过载和实现应用系统个性化的主要途径。实现上述推荐关系网络的基础是个性化推荐技术,它的研究涉及信息检索、数据挖掘、人工智能、社会学等众多领域。 “电子警务”是改变传统警务方式、提高警务工作水平的重要措施。电子警务的个性化是指以网络方式进行的警务活动个性化,就是公安部门如何利用电子警务平台,对警务活动进行再造,通过开放式网络环境向警员提供个性化的办公流程和服务。在电子警务的个性化实现中,个性化推荐技术是基本的关键技术之一。本论文围绕开放式环境下的个性化推荐技术、模型及算法展开调研,对个性化推荐技术、特别是现有推荐系统的社区结构发现技术进行了系统的分析和比较。针对用户行为的刻画与兴趣获取、相似用户查找、个性化推荐策略、和社区自组织等问题,进行了分析和研究。在此基础上,基于协同过滤技术,提出了一种基于多级智能代理的自组织互惠社区构建及推荐算法,来完成动态个性化推荐,并对该算法的有效性和鲁棒性进行了验证。最后,结合公安数据种类繁多,各警种对公安数据的兴趣参差不齐的特点,将上述个性化动态推荐技术的研究结果应用于电子警务领域,以减少警务人员信息过载的问题。并且针对某市电子警务现状,开发了基于学习社区监控和动态个性化数据资源推荐的电子警务查询系统,实现相同兴趣警员的自组织,并提供资源推荐、资源共享、协同交流等功能,帮助警员有效共享资源和经验。
[Abstract]:With the rapid development of modern communication technology, Internet technology and multimedia technology, users can quickly obtain various forms of information resources. It is also inevitable to bring about the problem of "information overload" (Information Overload). Internet media users or communicators have formed a new type of recommendation relationship network through their preference for resources, social relations, task cooperation and so on. Providing interested information for this kind of recommendation relation network is the main way to reduce information overload and realize the individualization of application system. The implementation of the recommendation relationship network is based on personalized recommendation technology, which involves many fields, such as information retrieval, data mining, artificial intelligence, sociology and so on. Electronic policing is an important measure to change the traditional policing mode and improve the police work level. The individualization of electronic policing refers to the individualization of police activities carried out by the network, that is, how the public security departments make use of the electronic policing platform to reconstruct the police activities. Through the open network environment to provide police officers with personalized office processes and services. In the personalized implementation of electronic policing, personalized recommendation technology is one of the basic key technologies. In this paper, the personalized recommendation technology, model and algorithm in open environment are investigated, and the personalized recommendation technology, especially the community structure discovery technology of the existing recommendation system, is systematically analyzed and compared. This paper analyzes and studies the characterization and interest acquisition of user behavior, similar user search, personalized recommendation strategy, and community self-organization. On this basis, based on collaborative filtering technology, a self-organizing mutually beneficial community construction and recommendation algorithm based on multi-level intelligent agent is proposed to complete the dynamic personalized recommendation, and the effectiveness and robustness of the algorithm are verified. Finally, combined with the characteristics of various kinds of public security data and uneven interest in public security data, the research results of the above personalized dynamic recommendation technology are applied to the field of electronic policing in order to reduce the problem of police personnel information overload. According to the present situation of electronic policing in a city, an electronic policing query system based on learning community monitoring and dynamic personalized data resource recommendation is developed to realize the self-organization of police officers of the same interest, and to provide resource recommendation and resource sharing. Collaborative communication and other functions to help police officers effectively share resources and experience.
【学位授予单位】：解放军信息工程大学
【学位级别】：硕士
【学位授予年份】：2008
【分类号】：D631.1

【参考文献】