面向在线微博网络的相似用户发现方法研究

发布时间：2019-01-07 10:07

【摘要】：随着互联网的盛行,主流的在线社交媒体取得了突飞猛进的发展,而微博在这些在线社交媒体中表现出了更加突出的发展态势。微博已经吸引了海量的用户,人们在微博上分享自己的观点、爱好、思想等信息,微博对社会的发展产生了巨大的影响。近年来国内外研究者基于在线社交媒体数据做了全方位的研究,如社区发现、商品推荐、相似用户推荐等。本文面向微博,开展微博用户个体的话题挖掘和相似用户挖掘的理论和方法研究。具体的研究内容包括如下三个方面:(1)根据微博动态社交的特点,以时间片划分用户的微博信息,挖掘用户的动态话题,该方法可以更好地体现微博用户话题的动态性,可以使相似用户的挖掘结果更加精准。(2)在全面分析微博网络用户特点的基础上,从三个方面计算用户之间的相似度:(1)用户背景属性的相似度;(2)用户动态话题的相似度;(3)用户社交关系的相似度。分别计算出三个方面的相似度后,最后通过加权的方式计算出两个微博用户的相似度,最后计算出特定用户与爬取出的微博用户集的相似度,根据相似度的大小生成降序排名队列,将排名靠前的Top-k个用户推荐给特定用户。(3)为了更加直观的展示并实现本文的研究成果,本文根据前两部分研究内容,设计并实现了基于微博的相似用户挖掘的原型系统,该原型系统主要包括动态话题挖掘以及相似用户挖掘等功能。本文涉及的研究内容,能够帮助用户发现相似用户可以使用户拓宽自己的人际关系,寻找到自己更感兴趣的话题,使其发现自身在社会网络中扮演的角色,可以为社区的发现做数据的支撑。
[Abstract]:With the popularity of the Internet, mainstream online social media have made rapid progress, and Weibo has shown a more prominent development trend in these online social media. Weibo has attracted a large number of users, people share their views, hobbies, ideas and other information on Weibo, Weibo has a great impact on the development of society. In recent years, researchers at home and abroad based on online social media data to do a full range of research, such as community discovery, commodity recommendations, similar user recommendations. This paper aims at Weibo and studies the theory and method of topic mining and similar user mining. The specific research contents include the following three aspects: (1) according to the characteristics of Weibo's dynamic social interaction, the paper divides the user's Weibo information by time slice and excavates the user's dynamic topic. This method can better reflect the dynamic nature of the user's topic. It can make the mining results of similar users more accurate. (2) on the basis of analyzing the characteristics of Weibo network users, the similarity between users can be calculated from three aspects: (1) the similarity of user background attributes; (2) similarity of user's dynamic topic and (3) similarity of user's social relationship. After calculating the similarity of three aspects respectively, the similarity of two Weibo users is calculated by weighted method, and the similarity between specific user and the user set taken out by crawling is calculated. According to the size of similarity, the descending rank queue is generated, and the top Top-k users are recommended to specific users. (3) in order to more intuitively display and realize the research results of this paper, this paper according to the first two parts of the research content, The prototype system of similar user mining based on Weibo is designed and implemented. The prototype system mainly includes dynamic topic mining and similar user mining. The research in this paper can help users to discover that similar users can broaden their interpersonal relationships, find topics of more interest to themselves, and make them discover their own role in social networks. Can do data support for community discovery.
【学位授予单位】：中国矿业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.1;TP393.092

【参考文献】