导　　师： 廖开际

学科专业： 1201

授予学位： 硕士

作　　者： ;

机构地区： 华南理工大学

摘　　要： 网络改变了传统的文献资源利用模式，也拓宽了人与人之间的交往空间，促进人们之间对于资源的交流，出现了为网络大众用户提供和获取资源的服务平台。在Web2.0时代，网络用户既是网站内容的使用者，也是网站内容的生产者，许多用户对相同资源的访问行为可以看作是大众行为。但随着网络信息和用户的增多，也为网络用户带来信息过载和认知负担，用户难以获取自己所感兴趣的信息，也难以结识相同兴趣的其他用户。这种局面已经成为社会网络服务提供商的经营困境之一，如何通过主动推荐为用户提供真正所需的服务也已经成为研究的焦点。 基于大众行为互联网服务发展迅速，已经融入到大部分网民的生活中。以社会学中的大众行为理论基础，从系统平台特点的角度分析了网络大众行为的六种主要形式：社会网络服务SNS；博客；大众评论；维基百科；论坛；微博。构建了大众行为下的社会网络结构图，并根据社会网络类型将大众行为下的社会网络分为用户间无向的社会网络、用户间有向的社会网络和以信息为中介构建社会网络三类。 传统社会网络研究主要是采用抽样调查方法获取实验数据。随着社会网络服务的盛行，网民的实际人际网络与网络上虚拟的社会网络重复度提升，采用网络挖掘的方法来获取用户的社会网络变得可行。选取大众服务网站豆瓣网作为实验数据来源，用户在豆瓣网上访问资源的行为属于大众行为，用户之间的关注关系属于社会网络。用户的兴趣特征可以从用户的好友、用户访问的资源及用户所参加的群组三个方面进行分析，用户-资源-群组三要素构成了大众行为下社会网络的信息传播机制。根据用户访问资源的行为记录，通过资源标签为中介分析用户的兴趣偏好，在此基础上采用交集的计算方法计算用户兴趣相似度，根据相似度高低向用户推荐同趣的其他用户。根据资源的标签和用户兴趣偏好，向用户推荐其感兴趣的资源。根据用户所参加的群组中的其他用户所在群组，向用户推荐其可能感兴趣的群组，提升用户活跃度。此外，群组主题与资源内容存在相关度，对某个资源感兴趣的用户，可以推荐其加入相关群组。研究用户兴趣对社会网络的影响，进而研究向用户推荐兴趣相似用户和用户所感兴趣的资源和群组，提升系统的服务质量。 The Internet has changed the traditional literature resource utilization pattern, has alsopromoved interaction between people, and promoted exchanges of resources, bringing theemergence of service platform for network public users accessing resources.Netizens are bothusers of the site content and the producer of the website content in Web2.0era, the behaviorof Netizens accessing same resources can be seen as mass behavior. Internet services based onmass behavior developes rapidly, which have been incorporated into life of the majority ofNetizens. From the perspective of mass behavior theory in sociology, six forms of Internetmass behavior are analyzed according to the system platform characteristics,that’s SocialNetwork Service,Blog, Public comment,Wiki, BBS and MicroBlog. The social networkstructure is classed in No to the social network, Have to social network and social networkconstructed by information intermediary. Along with the prevalence of social network service, interpersonal relationships networkand virtual social network repeatedly, which make the method of network mining method toget user's social network become feasible. Take public service website Douban/(www.douban.com/) as experimental data sources, behaviour of accessing resources online ismass behavior, and the relationship between users is social network. This is paper focused onuser service recommendation on the basis of user interest and its social network. User interestcould be analyzed from friends, access behavior and group joined, and the informationcommunication mechanism in mass behavior is formed by the user, resource and group.Labels of resource accessed by user reflect its interst. Resources are recommended to useraccording to its labels and user interst. Groups are recommended to user according to groupsof uers who are in the groups which the user joined. In addition, there is correlation betweentheme of group and the content of resource. Related groups could be recommended to the userif it’s interested in the resources. Research on the impact of user interst to social network, andthen recommend similar interst user, interested resources and interested groups to the user,which could improve service quality of social network service.

关 键 词： 大众行为 社会网络 用户兴趣 服务推荐

分 类 号： [F49]

领　　域： [经济管理]