基于社交网络信息的健康话题关注度预测方法研究

发布时间：2018-06-13 03:34

本文选题：话题发现 + 关注度预测　；参考：《北京邮电大学》2016年硕士论文

【摘要】：社交网络平台是伴随着移动互联网技术的发展而迅速普及起来的一种全新的信息交流方式,微博作为社交网络中普及范围最广的一种形式,在使用人数、用户活跃度等方面都已经远远超越了其他社交网络服务平台,已经成为一个有着充足用户资源的信息来源。随着人们对自身健康情况的重视和健康观念的转变,越来越多的人在微博平台上发布和交流与健康话题相关的信息,参与到健康话题的相关讨论中。因此,微博服务平台已经成为一个重要而丰富的健康信息数据来源。从这个丰富的信息来源平台上获取有关人们健康话题的信息和数据,对民众保持自身健康状况具有重要的意义,也有利于医学工作者提高相关工作的主动性和及时性,对其开展疾病监控和预防工作提供参考性的建议。因此本文针对微博平台上的健康话题发现和关注度预测方法进行了研究。根据调研,本文选择新浪微博作为社交网络信息来源。通过分析新浪微博平台的特点,研究了针对新浪微博平台获取微博数据的方法,并对健康话题关注度趋势预测所涉及的关键技术进行了调研和研究。在对微博文本特点、话题关注度发展及话题发现与追踪(Topic detection and tracking,TDT)技术的研究和学习的基础上,本文设计并实现了一个针对新浪微博数据的健康话题关注度预测系统。系统利用网络爬虫与新浪微博平台开放API相结合的技术手段分两个时间段采集了 75天内共170828条新浪微博数据作为分析对象。在对原始数据进行处理后,选用SVM算法从中筛选出健康领域相关的微博文本。对于筛选后的文本,使用Single-Pass算法进行文本聚类,完成话题抽取。本文定义了健康话题关注度计算方法,并据此在话题发现结果的基础上建立了马尔科夫预测模型完成对话题关注度趋势的预测。
[Abstract]:With the rapid development of mobile Internet technology, social network platform is a new way of information exchange. As the most widely used form of social network, Weibo is widely used. User activity has been far beyond other social networking platforms, and has become a source of information with sufficient user resources. With the attention of people to their own health situation and the change of health concept, more and more people publish and exchange information related to health topic on Weibo platform, and participate in the relevant discussion of health topic. Therefore, Weibo service platform has become an important and rich health information data source. Getting information and data on people's health topics from this rich information source platform is of great significance to the people to maintain their own health status, and also conducive to the medical workers to improve the initiative and timeliness of related work. To carry out disease control and prevention of the work to provide reference advice. Therefore, the methods of health topic discovery and attention prediction on Weibo platform are studied in this paper. According to the investigation, this article chooses Sina Weibo as the social network information source. By analyzing the characteristics of Sina Weibo platform, the method of obtaining Weibo data from Sina Weibo platform is studied, and the key techniques of predicting the trend of health topic concern are investigated and studied. On the basis of the research and study on the characteristics of Weibo text, the development of topic attention and topic discovery and tracking, this paper designs and implements a health topic concern prediction system for Sina Weibo data. The system uses the technology of web crawler and open API of Sina Weibo platform to collect 170828 Sina Weibo data in 75 days. After processing the raw data, the Weibo text related to health domain is selected by using SVM algorithm. For filtered text, Single-Pass algorithm is used for text clustering to complete topic extraction. Based on the results of topic discovery, a Markov prediction model is established to predict the trend of topic concern.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.1;TP393.092

【相似文献】