群体智慧在社交媒体中的应用研究

发布时间：2018-04-01 05:00

本文选题：群体智慧　切入点：社交媒体　出处：《大连理工大学》2014年硕士论文

【摘要】：社交媒体已逐渐成为人们生活中分享、交流、互动的新平台,其中,人们根据不同的兴趣、话题等凝聚形成多种群体。在社交群体的讨论交流过程中将产生十分庞大的信息量,这些信息具有杂乱、短周期性的特点。因此,从大量信息中获取高质量的信息,为用户提供更完整的信息和服务更加受到关注。社交群体存在并体现了群体智慧,运用群体智慧的思想和技术,分析社交群体的信息,能够更好的为用户提供服务也愈发重要。本文主要从群体智慧的角度,针对新浪微博中特定社交群体的文本信息进行研究,分别将群体智慧的思想应用到垃圾识别问题和排名算法两个问题中,为群体中的用户浏览微博提供方便。 (1)垃圾微博随着社交网络的发展日益增多,本文主要针对新浪话题,结合群体智慧的思想,从无监督学习的方法考虑,提出一种基于随机游走聚类的垃圾微博自动识别模型。该模型针对微博话题,根据微博间的相似度构建微博关系网络,通过随机游走聚类算法对微博进行聚类,从而划分多个微博群体,最后针对聚类得到的独立个体或小群体利用垃圾微博的显性特征识别过滤垃圾微博。实验证明,该模型可以有效地识别垃圾微博,且效果优于传统的监督学习方法,尤其是召回率方面表现明显,可为用户过滤微博中的垃圾内容,提高用户浏览效率。 (2)在对新浪微博的电影微吧群体的研究中,本文主要结合群体智慧的思想,提出了一种基于蚁群算法的排名模型(ACOR),该模型根据群体中用户的偏好以及电影的热度对电影进行综合排名。同时,该模型还考虑了微博信息中流露出的情感因素,通过分析和把握用户对电影的情感倾向,计算其情感值。最后,根据群体微博计算的情感积累值对热议的电影进行排名,实现了利用群体微博信息对电影的排名。该排名更加符合用户的偏好,并且具有一定的实时性,符合微博实时性强的特点,可以有效地为用户提供电影相关信息。
[Abstract]:Social media has gradually become a new platform for people to share, communicate, interact with each other in their daily lives, in which people form multiple groups according to their different interests, topics, etc. In the process of discussion and communication among social groups, there will be a very large amount of information. This information is cluttered and short-cyclical. Therefore, getting high quality information from a large amount of information, providing users with more complete information and services, is of greater concern. Social groups exist and embody group wisdom. It is more and more important to analyze the information of social groups by using the ideas and techniques of group intelligence to provide better service to users. From the angle of group intelligence, this paper studies the text information of specific social groups in Sina Weibo, and applies the thought of group intelligence to the problem of garbage recognition and ranking algorithm, respectively. For the group of users to browse Weibo to provide convenience. With the increasing development of social networks, this paper focuses on the topic of Sina, combines the thought of group intelligence, and considers the method of unsupervised learning. This paper presents an automatic identification model of garbage Weibo based on random walk clustering. According to the similarity of Weibo, the model is used to set up a relationship network between Weibo, and to cluster Weibo by random walk clustering algorithm. Finally, according to the dominant characteristics of the garbage Weibo, the model can be used to identify and filter the garbage Weibo. Experimental results show that the model can effectively identify the garbage Weibo. And the effect is better than the traditional supervised learning method, especially the recall rate is obvious, which can filter the spam content of Weibo for users and improve the efficiency of browsing. 2) in the study of the group of Sina Weibo's movie micro-bar, this paper mainly combines the thoughts of group wisdom. In this paper, a ranking model based on ant colony algorithm (ant colony algorithm) is proposed. The model ranks films synthetically according to the preferences of users in the group and the heat of films. At the same time, the model also takes into account the emotional factors revealed in Weibo's information. By analyzing and grasping the emotional tendency of the users to the film, we calculate the emotional value. Finally, according to the emotional accumulation value calculated by Weibo, we rank the hot films. This ranking is more in line with users' preferences, and has the characteristics of real-time and strong real-time, which can effectively provide users with motion-related information.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092

【相似文献】