基于用户活跃度和热门话题的微博社区推荐技术研究

发布时间：2018-10-22 17:26

【摘要】：微博为人们提供了一个广泛的平台，使他们能够在自己所在的虚拟社区中随时随地的分享信息，吸引了越来越多的用户，其中具有代表性的微博服务有Twitter和中国的新浪微博。微博中的用户关系大多是不对称的，因为一个用户可以有很多的粉丝，，但是这个用户并不需要反过去关注这些粉丝，用户有充分的主动权来选择和决定自己的行为而不是被动地做出行动。正是由于这种不对称关系，微博成为广播通信的媒介，其中信息在多个节点交互形成的大规模网络中传播。然而在这个浩如烟海的信息社区中如何快速、高效地获取到最有价值的信息是至关重要的。微博社区推荐可以看作是一定时期内针对更新信息的过滤和推荐，将获取到的最新的、对用户最有用的信息推荐给用户。目前很多国内外学者就这个方面提出了微博社区模型和微博社区推荐算法来提高社区推荐的精准度，并且取得了一定的工作成果。但随着网络技术的不断发展和社交工具的不断更新，人们表达自己观点的方式和平台呈现多样性，社交方式也在逐渐变化，这使得以往传统的微博社区推荐方法不足以满足新的社交网络的需求。基于上述背景，本文介绍了微博社区的基本理论，分析了现有的微博社区推荐技术，并在此基础上提出了基于用户活跃度和热门话题的微博社区推荐方法。论文的主要研究内容如下： (1)以已有的Web社区模型为基础，分析了微博社区内用户之间的行为关系、社区通信的范围并构建了行为模型和通信过程模型，且对微博社区的特征进行了分析。 (2)对微博用户进行了活跃度基础分析，给出了推文数量和用户数量关系、推文数量和粉丝数量关系。 (3)分析了用户的兴趣度并讨论了用户之间的关注关系，以此为依据，研究了微博热门话题检测过程中时间因素的重要性，并通过改进经典算法PageRank提出了一个考虑时间因素的热门话题检测算法。 (4)提出了基于用户活跃度与热门话题的微博社区推荐算法，从“活跃度”和“热门”这两个方面将微博社区中的活跃用户及社区中讨论的热门话题推荐给社区中的新用户，以提高这些新用户的检索速度和效率，又快又准的获取信息。 (5)本文对提出的方法进行实验验证，通过真实数据集上的实验结果来验证所提方法的高效性。
[Abstract]:Weibo provides a wide platform for people to share information anytime and anywhere in their virtual community, attracting more and more users, including Twitter and Chinese Sina Weibo. Most of the user relationships in Weibo are asymmetrical, because a user can have a lot of fans, but that user doesn't need to focus on these fans in the past. Users have sufficient initiative to choose and determine their own actions rather than passively. Because of this asymmetry, Weibo became the medium of broadcast communication, in which information spread in a large scale network formed by the interaction of multiple nodes. However, how to get the most valuable information quickly and efficiently is very important in this vast information community. Weibo's community recommendation can be regarded as filtering and recommending the updated information in a certain period of time and recommending the latest and most useful information to the user. At present, many scholars at home and abroad put forward Weibo community model and Weibo community recommendation algorithm to improve the accuracy of community recommendation, and achieved certain results. However, with the continuous development of network technology and the continuous updating of social tools, the ways and platforms of people expressing their views are diverse, and the social ways are also gradually changing. This makes the traditional Weibo community recommendation method not enough to meet the needs of new social networks. Based on the above background, this paper introduces the basic theory of Weibo community, analyzes the existing community recommendation technology of Weibo, and puts forward a community recommendation method based on user activity and popular topic. The main contents of this paper are as follows: (1) based on the existing Web community model, this paper analyzes the behavior relationship among users in Weibo community, the scope of community communication, and constructs the behavior model and communication process model. And the characteristics of Weibo community were analyzed. (2) the basic analysis of the activity degree of Weibo users was made, and the relationship between the number of tweets and the number of users was given. The relationship between the number of tweets and the number of followers. (3) based on the analysis of the interest of users and the discussion of the relationship of concern among users, the importance of time factors in the process of testing hot topics by Weibo is studied. And by improving the classical algorithm PageRank proposed a hot topic detection algorithm considering time factors. (4) proposed Weibo community recommendation algorithm based on user activity and hot topics. From the aspects of "activity" and "popularity", the active users in Weibo community and the hot topics discussed in the community are recommended to the new users in the community in order to improve the search speed and efficiency of these new users. It is fast and accurate to obtain information. (5) the proposed method is verified experimentally, and the effectiveness of the proposed method is verified by the experimental results on the real data set.
【学位授予单位】：兰州交通大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092;TP391.3

【相似文献】