基于时序信息的话题动态演化模型及追踪算法

摘要

针对舆情话题演化的时序特性、衍生特性和话题漂移现象，在分析话题演化特性的基础上构建了基于时间片的话题动态演化模型，并设计了话题追踪算法，将网页内容与链接关系归一化来计算话题相关度。实验表明，该算法能够有效应对话题漂移和话题衍生效应，显著提高了话题的追踪效果。%The timing characteristics , derivative features and topic drift in evolution of public opinion topic were analyzed . The dynamic evolution model was established based on the analysis .The topic tracking algorithm was designed .The relevance between topics was computed when web content was combined with links analysis .The experimental results show that the algo-rithm can effectively deal with topics drift and derivative effects and significantly improves the effectiveness of integrated indicators of topic tracking .