面向舆情监控的智能化自然语言处理算法设计  

Design of intelligent natural language processing algorithm for public opinion monitoring

作　　者：罗涛 谢凤祥 李光华 LUO Tao;XIE Fengxiang;LI Guanghua(National Energy Group Dadu River Basin Hydropower Development Co.,Ltd.,Chengdu 610000,China)

机构地区：[1]国能大渡河流域水电开发有限公司,四川成都610000

出　　处：《电子设计工程》2023年第21期114-118,共5页Electronic Design Engineering

摘　　要：针对当前网络舆情分析效率较低且准确度不足的问题,开展了基于自然语言处理的智能舆情监控算法设计研究。该研究设计构建了网络舆情智能监控技术框架,其结构包括数据源层、数据采集层、数据处理层与舆情分析应用层。同时在此基础上提出一种融合多维注意力机制的智能舆情监控算法,通过利用网络爬虫技术生成了舆情语料库,采用CBOW词向量模型完成自然语言的结构化表示,并使用多维注意力机制网络挖掘舆情信息与舆情风险之间的内在关联,进而实现了对舆情风险等级的准确监控。仿真分析结果表明,所提出的CBOW模型在自然语言表示上计算速度更快、准确度也较高。而多维注意力机制网络在舆情风险等级预测上更为精确,且能够准确把握网络舆情动向,为企业运营提供指导。Aiming at the problem of low efficiency and accuracy of network public opinion analysis,this paper carries out the design and research of intelligent public opinion monitoring algorithm based on natural language processing.This paper designs and constructs the technical framework of intelligent monitoring of network public opinion,including four layers:data source layer,data acquisition layer,data processing layer and public opinion analysis application layer.On this basis,an intelligent public opinion monitoring algorithm integrating multi-dimensional attention mechanism is proposed.The public opinion corpus is generated by Web crawler technology,the CBOW word vector model is used to realize the structured representation of natural language,and the multi-dimensional attention mechanism is used to mine the internal relationship between public opinion information and public opinion risk,so as to realize the accurate monitoring of public opinion risk level.The simulation results show that the CBOW model proposed in this paper has faster calculation speed and higher accuracy in natural language representation,and the multi-dimensional attention mechanism network is more accurate in the prediction of public opinion risk level.It can accurately grasp the trend of network public opinion and provide guidance for enterprise operation.

关 键 词：舆情 爬虫 注意力机制 风险识别 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置] TN99[自动化与计算机技术—控制科学与工程]