基于互联网传播文本的地区环境形象评价方法  

An Evaluation Method of Regional Environmental Image Based on Internet Environmental News

作　　者：段长宇 胡裕民 赵志杰[1] 李晓亮 Duan Changyu;Hu Yumin;Zhao Zhijie;Li Xiaoliang(College of Environmental Science and Engineering,Peking University,Beijing 100871,China;Chinese Academy of Environmental Planning,Beijing 100012,China)

机构地区：[1]北京大学环境科学与工程学院,北京100871 [2]生态环境部环境规划院,北京100012

出　　处：《科技管理研究》2022年第19期244-250,共7页Science and Technology Management Research

基　　金：生态环境部环境规划院课题“城市环境规划管理”(2020A049)。

摘　　要：基于互联网新闻,明晰“地区环境形象”的概念,划分文体来源、情感极性和环境要素3种环境形象类别,利用文本分类、情感分析等自然语言处理手段,开发基于互联网文本信息的地区环境形象评价方法。人工标注环境语料库,对比3种学习算法,最终选取卷积神经网络算法训练评价模型。方法的分类效果较好,环境要素的微平均值(F1值)在0.8~0.9之间,情感分析的F1值在0.8以上,文体来源的F1值在0.9左右。该方法应用在长三角城市,可实时处理地区热点环境舆情、分析地区环境形象,为区域环境管理提供基础信息支持。Based on Internet news, the concept of regional environmental image is clarified, and three types of environmental image are divided: stylistic sources, emotional polarity and environmental elements. Using natural language processing methods such as text classification and sentiment analysis, a regional environmental image evaluation method based on Internet text information is developed. The environmental corpus was manually labeled,and three learning algorithms were compared. The convolutional neural network algorithm was finally selected to train the evaluation model. The classification effect of the method is good, the micro-average(F1 value) of environmental elements is between 0.8 and 0.9, the F1 value of sentiment analysis is above 0.8, and the F1 value of style source is about 0.9. This method is applied in the Yangtze River Delta cities, which can process regional environmental public opinion in real time, analyze the regional environmental image, and provide basic information support for regional environmental management.

关 键 词：环境新闻 环境形象 自然语言处理 生态环境大数据 环境管理 

分 类 号：X321[环境科学与工程—环境工程] G32[文化科学]