作　　者： ;

机构地区： 华南理工大学经济与贸易学院

出　　处： 《情报杂志》 2013年第12期42-46,11共6页

摘　　要： 随着微博应用的迅猛发展，互联网成为当今社会重要的舆论场。从海量微博信息中了解并预测出网民所关注话题的发展趋势成为一个具有挑战性的研究课题。提出基于意见领袖参与行为的微博话题热度预测模型，以意见领袖的影响力及其参与行为等指标，研究微博热点话题的发展趋势和特征，分析并验证了模型对不同话题类型的适用性，针对话题演化过程的阶段性变动特征，对模型进行了改进和完善，提出了基于动态系数的多元线性回归预测模型。实验结果表明，预测模型在不考虑外界因素刺激的情况下可以获得较好的预测结果。 With the rapid development of mieroblog, Interact has become an important public opinion field in today＇ s society. Predicting the trend of the microblog hot topics from the vast information becomes a challenging research concern. In this paper, we studied and put forward a hotness prediction model of microblog topics based on the participation of opinion leaders. Measurements with the influence and the participation of the opinion leader were regarded as the key indicators of the trend of the mieroblog topic hotness. The applicability of the model on different kinds of topics was also analyzed. Considering the periodic fluctuation characteristics of the topic evolution process, we improved the model and built a multiple linear regression prediction model based on dynamic coefficient. It was proved that without considering the stimulation of external factors, the hotness prediction model of microblog topics could yield desirable prediction.

关 键 词： 意见领袖 微博话题 热度预测 动态回归模型

分 类 号： [TP391.1]

领　　域： []