多源区域民生话题演化技术研究  

Research on the Evolution Technology of Livelihood Topics in District Areas from Multiple Data Resources

作　　者：张晓明[1] 申晴 王芳 赵培森 于占鲁 Zhang Xiaoming;Shen Qing;Wang Fang;Zhao Peisen;Yu Zhanlu(College of Information Engineering,Beijing Institute of Petrochemical Technology,Beijing 102617,China)

机构地区：[1]北京石油化工学院信息工程学院,北京102617

出　　处：《南京师大学报（自然科学版）》2023年第3期105-111,共7页Journal of Nanjing Normal University(Natural Science Edition)

基　　金：北京市优秀人才项目(ZZB2019005);北京市科技计划一般项目(KM202010017011).

摘　　要：民生一直是社会重点话题,近两年的疫情防控又为话题聚焦和演化注入了新的内容.本文基于大量区域化民生数据进行LDA模型的困惑度分析,证明多源文本话题比单源文本更全面.并进一步提出了民生话题演化技术框架,创新设计了热度演化率和关键词演化率的计算方法和实现算法.基于HTDI模型和关键词演化率,综合设计了民生话题演化指数LTEI.实验数据采集于北京大兴区的官方微博和百度贴吧.实验结果表明,TF-IDF模型比TextRank模型更合适计算关键词演化率;与HTDI指数相比,LTEI指数与实际话题演化趋势更加贴合,更适合用于区域民生话题演化分析.The topic of people’s livelihood has always been a key social issue.The epidemic prevention and control in the past two years has injected new content into the focus and evolution of the topic.Based on a large number of collected regional livelihood topic data,the perplexities as LDA model are analyzed to show that the LDA topics from the multiple source data are more comprehensive than the individuals.Then,a kind of technique framework of livelihood topic evolution is put forward firstly.Some new ideas of heat evolution rate(ER)and keyword ER are created with detail definition and concrete algorithms.Furthermore,based on the HTDI model and keyword ER,the comprehensive model as livelihood topic evolution index(LTEI)is designed for topic evolution process.The data set is collected online from official Weibo,Baidu Tieba mainly in Daxing District of Beijing.The experimental results show that the TD-IDF model is more suitable for keyword ER than TextRank model.Compared with HTDI,the LTEI is more consistent with the evolution trend of actual topics and is more suitable for the evolution of regional livelihood topics.

关 键 词：话题演化 民生 演化率 演化指数 新冠肺炎疫情 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]