社交网络用户影响力算法研究与实现

发布时间：2019-07-01 11:45

【摘要】：近些年来,社交网络的迅速发展对人们生活产生越来越深远的影响,社交网络因其方便、快捷、及时等特点吸引了大量用户。对社交网络用户进行深入而全面分析与挖掘,在舆情控制、信息传播、广告投放等领域有很大意义,因此成为热门研究内容。如今的社交网络呈现出复杂多样、数据量大等特点,如何基于这些特点,准确而高效地分析社交网络用户显得十分必要,这也是本文的主要研究内容。本文首先分析了以微博为代表的社交网络的特点,并在CASINO算法的基础上,考虑了微博话题下,社交网络用户的微博点赞数、转发数和评论数,提出了一种改进算法——TPURANK算法,并用该算法来分析用户的影响力指数,实验数据表明TPURANK算法的分析结果更具科学性和合理性。其次,本文在分析TPURANK算法的基础上,结合MapReduce编程模型,提出了一个TPURANK算法的并行化方案,我们在Hadoop平台上进行了实现,并在不同的条件下进行测试,实验结果表明,该算法与原有的CASINO算法相比,具有较好的集群性能和更快的执行速度,这有助于我们分析社交网络的海量数据,以适应社交网络用户数量越来越多、规模越来越大的特点。再次,在前面工作的基础上,我们设计并实现了社交网络分析系统,包括需求分析、总体设计、功能模块设计以及具体实现等工作,该系统具有社交网络用户影响力指数排名、依从性指数排名、用户相关信息查询等功能。最后,我们对社交网络分系统的计算结果进行了深入地分析和比较,发现社交用户的影响力指数与入链数、点赞数、转发数和评论数有关,这与实际情况是非常相符的,具有较高的参考价值,因此该系统非常适合对社交网络用户进行分析。此外,我们还对本论文的工作进行了全面的总结以及对未来的工作进行了展望。社交网络作为现代信息传播的重要平台之一,有着极高的研究价值和广阔的应用前景,深入分析社交网络用户,不仅可以帮助我们发现更具价值的信息,而且能推动社交网络的发展,这也是我们的研究工作的重要目标。
[Abstract]:In recent years, the rapid development of social networks has a more and more profound impact on people's lives. Social networks have attracted a large number of users because of their convenience, rapidity, timeliness and so on. The in-depth and comprehensive analysis and mining of social network users is of great significance in the fields of public opinion control, information dissemination, advertising and so on, so it has become a hot research content. Nowadays, social networks show many characteristics, such as complex and diverse, large amount of data and so on. Based on these characteristics, it is necessary to analyze social network users accurately and efficiently, which is also the main research content of this paper. In this paper, the characteristics of social network represented by Weibo are analyzed, and on the basis of CASINO algorithm, an improved algorithm, TPURANK algorithm, is proposed, and the analysis results of TPURANK algorithm are more scientific and reasonable, considering the Weibo topic, the number of Weibo points, the number of forwarding points and the number of comments, and the improved algorithm is used to analyze the influence index of users. The experimental data show that the analysis results of Weibo algorithm are more scientific and reasonable. Secondly, based on the analysis of TPURANK algorithm and MapReduce programming model, this paper proposes a parallelization scheme of TPURANK algorithm, which is implemented on Hadoop platform and tested under different conditions. the experimental results show that the algorithm has better cluster performance and faster execution speed than the original CASINO algorithm, which is helpful for us to analyze the massive data of social network. In order to adapt to the increasing number and scale of social network users. Thirdly, on the basis of the previous work, we design and implement a social network analysis system, including requirements analysis, overall design, functional module design and specific implementation. The system has the functions of social network user influence index ranking, compliance index ranking, user related information query and so on. Finally, we analyze and compare the calculation results of social network subsystem deeply, and find that the influence index of social users is related to the number of links, likes, forwarding and comments, which is very consistent with the actual situation and has high reference value, so the system is very suitable for the analysis of social network users. In addition, we also make a comprehensive summary of the work of this paper and look forward to the future work. As one of the important platforms of modern information dissemination, social network has high research value and broad application prospect. In-depth analysis of social network users can not only help us to find more valuable information, but also promote the development of social network, which is also an important goal of our research work.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP301.6;C912.3

【相似文献】