基于演化聚类的网络舆情数据挖掘系统设计

Design of network public opinion data mining system based on evolutionary clustering

为了提升网络舆情数据挖掘的召回率、精度和查准率,设计一个基于演化聚类的网络舆情数据挖掘系统。从综合需求、功能需求两方面进行深入分析,确定挖掘系统的实际要求。构建挖掘系统整体架构,采用垂直化结构分别设计支撑层、数据层、服务层、功能层,应用演化聚类建立舆情信息抓取关键词词库,计算随机演化初始向量,得到实验向量,分析不同维度的目标向量演化值,实现网络舆情抓取功能,确定适值函数实现信息挖掘。实验结果表明,设计系统的召回率能达到95%,精度能达到90%,且查准率接近99%,说明该系统的应用价值较高。

In order to improve the recall rate,precision and precision rate of network public opinion data mining,a network public opinion data mining system based on evolutionary clustering is designed.In⁃depth analysis from two aspects of comprehensive requirements and functional requirements,functional requirements and performance requirements to determine the actual requirements of the mining system.Build the overall architecture of the mining system,use the vertical structure to design the support layer,data layer,service layer and function layer respectively,apply evolutionary clustering to establish the keyword Thesaurus of public opinion information capture,calculate the initial vector of random evolution,get the experimental vector,analyze the evolution value of the target vector of different dimensions,realize the function of network public opinion capture,and determine the fitness function to realize information mining.The experimental results show that the recall rate of the designed system can reach 95%,the accuracy can reach 90%,and the precision rate is close to 99%,indicating that the application value of the system is high.