基于SVM的地震微博评价文本情感分类模型构建  

Construction of an SVM-based Emotion Classification Model forEarthquake-related Comments on Weibo

作　　者：杨万里 宋娟 任烨 YANG Wanli;SON Juan;REN Ye(Shanghai Earthquake Agency,Shanghai 200062,China;Shanghai Sheshan Monitoring Geophysical Obsarvatory and Research Station,Shanghai 200062,China)

机构地区：[1]上海市地震局,上海200062 [2]上海佘山地球物理国家野外科学观测研究站,上海200062

出　　处：《四川地震》2025年第2期46-51,共6页Earthquake Research in Sichuan

基　　金：上海佘山地球物理国家野外科学观测研究站重点课题(SSKP202301)。

摘　　要：为了获取地震灾后民众的情感变化情况,以关于2022年四川泸定Ms_(6).8地震的10万余条微博评论作为数据源,利用支持向量机(SVM)算法建立地震评价文本情感分类模型。通过多种核函数的模型比较,高斯核(g=0.3,C=10)时的SVM模型表现较好。通过此模型运算结果显示,泸定Ms_(6).8地震震后一个月内的微博评论具有阶段性特点,但均以正面评论为主。该方法可为地震舆情分析产品的开发打下基础,为有关部门舆情决策提供科学依据。To obtain the emotional changes of the public following an earthquake disaster,we used approximately 100000 comments from Weibo during the 2022 Luding M_(s)6.8 earthquake in Sichuan Province.Moreover,we established an emotional classification model for earthquake comments by using the support vector machine(SVM)algorithm.Furthermore,through comparing models with multiple kernel functions,we suggest that the SVM model under a Gaussian kernel(g=0.3,C=10)performs well.The calculation results of this model show that the Weibo comments within one month after the Luding M_(s)6.8 earthquake indicates phased characteristics highlighted by mainly positive comments.This method could serve as the foundation for the development of analysis products for public sentiment during earthquakes,which also provides a scientific basis for relevant departments'decision-making for public sentiment.

关 键 词：微博 支持向量机 地震 情感分类 2022年四川泸定M_(s)6.8地震 

分 类 号：C912.63[经济管理] P315.9[社会学]