基于微博的人物关系强度预测模型研究及实现

发布时间：2018-07-05 13:29

本文选题：新浪微博 + 用户关系　；参考：《西安电子科技大学》2014年硕士论文

【摘要】：随着信息技术产业的快速发展,宽带网络和高速移动网络的速度大幅提升,各种新型网络接入终端设备也变得越来越普及,极大地提高了人们通过网络进行沟通交流的速度和频率。当下,微博已成为网络思想交流的重要平台、网络舆情的高度聚集地,是人们交流思想、政府洞察民意的重要窗口。高用户活跃度和大量博文背后隐藏着重大的数据价值,通过代理数据,准确理解微博用户的交互行为、挖掘用户关系等数据中所蕴含的信息和影响具有重要的意义。本文基于微博平台对其所包含的用户关系进行分析和研究,主要研究内容和研究成果如下:1.通过对目前微博用户关系研究现状进行整理和分析,选定人物关系强度作为研究对象,从新浪微博的用户关注和互动关系出发,设计了人物关系强度预测模型。虚拟社交网络是现实社会社交网络的映射,其人物间的关系强度也具有相似的影响因素。因此,此模型通过对现实世界中人与人之间关系强度影响因素进行全面考虑,并将其迁移应用至新浪微博平台,抽取微博信息中反映用户间关系强度的可用信息,使用相似度、标准差等数学算法或概念将反映用户关系强度的可用信息转变为具体的数值信息,最终通过线性模型将多个影响因素综合从而进行定量化分析。2.实现了基于微博的人物关系强度预测系统。本系统中最重要的数据采集部分使用基于模拟登录和基于新浪微博API两种方式采集数据,实现了微博用户关系信息、用户个人资料信息和用户微博内容的自动提取。同时,综合使用上述两种数据采集方法,不仅避免了直接使用API的数据获取限制和用户未登录所造成的网页内容获取数据的不完整的问题,而且降低了大量数据分析和提取的工作量。3.人物关系强度预测系统根据设计的人物关系强度预测模型,对系统中数据采集部分获取的数据进行整理、分析和计算,预测人物关系强度,同时通过图形化界面展示人物关系强度。最后将某微博用户的预测模型结果与新浪微博人脉关系示例进行比较,证明了所设计的人物关系强度模型的有效性。本文所研究的人物关系强度能够对微博中的用户进行更准确的亲疏关系划分,基于本文的研究能够支持进一步使用社团分区算法进行更高准确度的好友推荐;支持舆情的精准发现,在舆情预警机制的使用中提高舆情预警的准确度;支持向用户推荐不同好友的隐私保护策略,帮助识别用户好友,同时保护用户隐私。
[Abstract]:With the rapid development of the information technology industry, the speed of broadband network and high speed mobile network has been greatly improved. All kinds of new network access terminal equipment have become more and more popular, which has greatly improved the speed and frequency of communication and communication through the network. At present, micro-blog has become an important platform for network thought communication, network public opinion The highly aggregated area is an important window for people to exchange ideas and the government's insight into the public opinion. High user activity and a large number of blog posts are hidden with significant data value. Through proxy data, it is important to understand the interactive behavior of micro-blog users and to excavate the information and influence in the data of user relations. The main research content and research results are as follows: 1. by sorting and analyzing the current research status of micro-blog user relations, the relationship strength of the characters is selected as the research object, and the relationship intensity of sina micro-blog is designed to predict the intensity of the relationship. Model. Virtual social network is the mapping of social social network in real society, and the relationship intensity of human and object has similar influence factors. Therefore, this model is fully considered by the influence factors of the relationship intensity between people in the real world, and applies its migration to the Sina micro-blog platform and extracts micro-blog information to reflect the users. Available information of relationship intensity, using mathematical algorithms or concepts such as similarity, standard deviation and other mathematical algorithms to convert the available information of user relationship strength into specific numerical information. Finally, a number of factors are synthesized by linear model and then quantificationally analyzed by.2.. In this system, a micro-blog based relationship strength prediction system is implemented. The most important data acquisition part uses two kinds of data acquisition based on analog logon and Sina micro-blog API. It realizes the micro-blog user relationship information, the user's personal information and the automatic extraction of the content of the user's micro-blog. At the same time, the comprehensive use of the above two data acquisition methods not only avoids the direct use of data acquisition limits for the use of the data acquisition methods. And the incompleteness of data obtained from the content of web pages caused by users not logged in, and reducing a large amount of data analysis and extraction of the workload.3. character relationship intensity prediction system based on the predicted model of the relationship strength of personage, the data collected in the data collection part of the system are collated, analyzed and calculated, and the figure is predicted. In the end, a micro-blog user's prediction model is compared with the Sina micro-blog relationship example, and the validity of the model is proved. The relationship strength studied in this paper can be more accurate and close to the users in the micro-blog. Relationship division, based on this study, it can support the further use of community partition algorithm for better friend recommendation, support the accurate discovery of public opinion, improve the accuracy of public opinion early warning in the use of public opinion early warning mechanism, support the privacy protection strategies of different friends to users, help identify the user friends, and protect the users. Protect the user's privacy.
【学位授予单位】：西安电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092

【参考文献】