基于流量经营平台的微博内容检测和追踪技术研究

发布时间：2018-04-29 22:31

本文选题：微博 + 传播模型　；参考：《北京邮电大学》2014年硕士论文

【摘要】：当今社会已经进入信息时代,随着互联网技术的发展,电信行业的市场竞争不断加剧,而这竞争不仅仅来自于同质行业的压力,也有其他异质企业的竞争,因此,运营商能否快速开展流量经营的工作成为当前工作的首要目标。从现状来看,流量业务已经明显的替代了传统的语音业务,成为运营商盈利的主要增长点,在提高运营商收入方面起着举足轻重的作用。某移动运营商搭建了国内首个流量经营平台,借助平台收集移动终端用户信息,来实现运营商深度了解客户需求,做到用户终端、网络、业务、内容可识别的程度,确保流量业务营销的精细化策划与执行。微博作为当下最流行的社交平台、舆论监督平台和网络新媒体,蕴含着海量用户信息。但是,流量经营平台中并没有一个模块能够对微博用户的具体信息进行分析,因此,本文基于平台原有架构和业务功能,针对微博在话题检测和追踪以及预警技术方面进行了研究并给出了可行的设计方案：首先利用流量经营平台的数据采集功能获取新浪微博转发关系和内容数据,结合文本信息的“质心”和转发结构的“中心性”过滤内容；然后提取微博短文本特征,通过区分噪声微博提出一种基于加权的增量聚类内容检测算法,通过计算微博内容中核心词汇的权重之和来判断微博是否归于话题簇中,该算法提高热点话题的检测效率；找到热点话题后锁定关键人物,分析用户转发行为,预测微博转发数量,实现异常话题的预警机制。
[Abstract]:Nowadays, the society has entered the information age, with the development of Internet technology, the market competition of the telecommunication industry has been intensified, and this competition not only comes from the pressure of the homogeneous industry, but also from the competition of other heterogeneous enterprises. Whether the operator can carry out the work of flow management quickly becomes the primary target of the current work. From the current situation, traffic service has obviously replaced the traditional voice service, become the main growth point of the operator profit, and play a pivotal role in improving the operator's income. A mobile operator has built the first domestic traffic management platform, with the help of the platform to collect mobile terminal user information, to achieve the operator to understand customer needs in depth, to achieve user terminal, network, business, content can be recognized. Ensure that flow business marketing is carefully planned and executed. Weibo, as the most popular social platform, public opinion supervision platform and network new media, contains massive user information. However, there is not a single module in the traffic management platform that can analyze the specific information of Weibo users. Therefore, this paper is based on the original framework and business functions of the platform. According to Weibo's research on topic detection and tracking and early warning technology, a feasible design scheme is given. Firstly, the data collection function of traffic management platform is used to obtain the relaying relationship and content data of Sina Weibo. Combining the "centroid" of text information with the "centrality" of the forwarding structure, and then extracting Weibo's short text features, an incremental clustering content detection algorithm based on weighted algorithm is proposed by differentiating noise Weibo. By calculating the sum of the weights of the core words in Weibo's content, the algorithm can improve the efficiency of detecting hot topics, lock key people after finding hot topics, and analyze the forwarding behavior of users. Forecast Weibo forward quantity, realize the early warning mechanism of abnormal topic.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.092

【相似文献】