摘要
[目的/意义]实现政府非正式文本的自动处理,提高公众留言处理的效率,提升政府形象。[方法/过程]以某市政媒融合问政平台公众留言为实验数据,采用基于Word2Vec和支持向量机的分类算法实现公众留言的自动分类,形成分类标签。根据一定的映射规则,形成标签集、职责集、职能部门集之间的对应,最终实现公众留言到政府职能部门的匹配。[结果/结论]总体分类准确率达到86.18%,实现了较好的分类效果,构建的政媒融合问政平台非正式文本的分类匹配模型具备合理性和科学性,使得非正式文本分类匹配问题得到了较好的解决。[局限]匹配环节未能具体实现以验证匹配效果,未进行多种分类算法的效果对比。
[Purpose/significance]This study aims to improve the efficiency of public message processing and enhance the government image via the automatic processing of government informal text.[Method/process]Taking the public message of the political platform of a local municipal government and the media as the experimental data,the classification algorithm based on Word2 Vec and support vector machine is adopted to realize the automatic classification of public message and form the classification labels.According to certain mapping rules,mapping among label sets,responsibility sets and functional departments are implemented,and finally match public messages with government functional departments.[Result/conclusion]The overall classification accuracy rate can reach 86.18%,achieving a good classification effect.The informal text classification and matching model for the political media integration platform is reasonable and scientific,and the problem of informal text classification and matching is solved well.[Limitations]The matching effect through empirical research fail to verify for lack of functional departments in reply message,and the effect comparison of various classification algorithms was not conducted.
出处
《情报理论与实践》
CSSCI
北大核心
2020年第6期156-161,148,共7页
Information Studies:Theory & Application
基金
国家社会科学基金重点项目“基于全生命周期的政府开放数据整合利用机制与模式研究”的成果,项目编号:17ATQ006。
关键词
政媒融合问政平台
文本分类
支持向量机
匹配模型
political media integration platform
text classification
support vector machine
matching model