基于主题图谱的网络舆情特征演化及其可视化分析

Visual Analysis of Network Public Opinion Feature Evolution Based on Topic Map

【目的/意义】旨在从网络舆情用户信息及文本内容视角出发,构建不同维度的网络舆情主题图谱,结合主题图谱对网络舆情进行特征演化及可视化分析,为舆情管理提供参考。【方法/过程】本文以实体抽取和关系构建技术为基础,构建了网络舆情主题图谱模型,并以"台风利奇马"事件为例,建立了三个不同维度的主题图谱,结合用户和文本等多维度微观数据,对网络舆情特征演化进行分析。【结果/结论】在该事件中,用户影响力节点具备多元化、相关性、官方主导性等特点;网络舆情演化对应台风事件发展存在一定的滞后性;PC终端存在传播媒介种类少、发博数量多且用户集中等特点,移动终端存在传播媒介种类多、发博数量少且用户分布均匀等特点。【创新/局限】本文借助主题图谱,构建了网络舆情用户节点和文本节点及其关联关系,从用户、账户、内容三个维度系统且全面的展示了网络舆情特征的演化规律。

【Purpose/significance】The topic map is used to analyze the network public opinion from multiple dimensions.【Method/process】This paper proposes a model for constructing the topic map of network public opinion.Firstly,the user entity,account entity,content entity and other entities are extracted from the social media data,and then the relationship between the entities is established by the relationship of transfer and praise,time flow,member information and media.Based on this model,the network public opinion topic map is constructed,and the characteristics of network public opinion can be visualized by the constructed map Analysis.The model is applied to the"Tphoon Lichma"event,and the user influence topic map,the network public opinion evolution topic map and the user topic map are constructed.Based on the map and micro data,the network public opinion feature evolution visualization analysis is carried out for the case event.【Result/conclusion】In the"Tphoon Lichma"network public opinion event,the user influence nodes have the characteristics of diversification,relevance and official dominance;the evolution of network public opinion has a certain lag in the development of typhoon events;PC terminal has the characteristics of less media types,more blogs and concentrated users,while mobile terminals have the characteristics of more media types,less blog quantity and uniform distribution of users.【Innovation/limitation】Based on the topic map,this paper proposes a new analysis method of network public opinion characteristics evolution,which provides reference for public opinion management.