Abstract: 【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 analy⁃ sis 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 cer⁃ tain lag in the development of typhoon events; PC terminal has the characteristics of less media types, more blogs and concentrated us⁃ ers, while mobile terminals have the characteristics of more media types, less blog quantity and uniform distribution of users.【Innova⁃ tion/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.