基于中文微博的话题趋势预测系统的设计与实现

发布时间：2018-05-30 12:55

本文选题：社交网络 + 话题趋势预测　；参考：《北京邮电大学》2014年硕士论文

【摘要】：近些年来社交网络得到了快速的发展,其独有的特性使得用户发布消息更加容易,信息传播更加快捷,然而社交网络的发展也对于商业机构的管理者和政府机构的决策者,提出了一定的挑战。这些机构的管理者和决策者需要随时并且快速的应对社交网络上面的热点话题及相关事件,因为这些事件有时会对相关机构产生非常现实且巨大的风险。本论文将以新浪微博为实际的例子探索如何在社交网络中对于话题趋势做出预测。在本文中,首先给出了社交网络中热点话题趋势预测的形式化定义和一个话题趋势预测模型,然后提出了一种针对时间序列数据进行特征抽取和构造的新方法,基于提出的新方法构造了三个相互补充的特征子集,最后将这三个特征子集线性组合起来做为支持向量机SVM的输入向量进行话题趋势的预测。为了验证预测模型的性能,本论文采集了一个高质量的新浪微博数据集。基于获取到的数据,对预测模型进行了完整的测试,取得了良好的效果。论文首先介绍了近些年来社交网络的发展及社交网络的相关研究领域；在第二章中论文提出了基于SVM的话题趋势预测模型,并且详细的给出了特征向量的构造方式；在第三章论文给出了爬虫框架的详细设计与实现,然后描述了话题趋势预测系统内部各个子系统的详细设计与实现；论文在第四章对于整个系统进行了详细的功能测试和性能测试,测试表明整个系统达到了预期的设计目标;在最后一章,论文对未来的工作进行了展望,并总结了作者在研究生期间的所有工作和成果。
[Abstract]:Social networks have developed rapidly in recent years. Their unique features make it easier for users to publish news and faster information dissemination. However, the development of social networks also poses a challenge to managers and decision-makers in business organizations. The managers and decision-makers of these institutions need to be ready and quick. We should deal with hot topics and related events on social networks quickly, because these events can sometimes pose a very real and huge risk to related institutions.
This paper will take Sina and micro-blog as an example to explore how to predict topic trends in social networks. In this paper, the formal definition of trend prediction and a topic trend prediction model of hot topics in social networks are given, and then a needle for feature extraction and construction of time series data is proposed. The new method is based on the proposed new method to construct three complementary feature subsets. Finally, the three feature subsets are combined linearly as the input vector of the support vector machine SVM to predict the topic trend. In order to verify the performance of the prediction model, a high quality Sina micro-blog data set is collected in this paper. The data have been tested on the prediction model and achieved good results.
This paper first introduces the development of social networks and the related research fields of social networks in recent years. In the second chapter, the thesis puts forward the prediction model of topic trend based on SVM, and gives a detailed description of the construction of feature vectors. In the third chapter, the detailed design and implementation of the crawler framework are given, and then the topic is described. The detailed design and implementation of each subsystem in the trend prediction system; in the fourth chapter, the detailed functional test and performance test are carried out for the whole system, and the test shows that the whole system has achieved the expected design goal. In the last chapter, the paper looks forward to the future work and summarizes the author during the graduate period. All the work and results.
【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;TP391.1

【相似文献】