[Purpose/significance] The spread of online public opinion in social media intensifies, gathers and ex? plodes social contradictions, which generates a huge impact on social stability and public values. This paper aims to ex? plore the evolution mode and law of public opinion by constructing the social network public opinion evolution map and analyzing the spatio-temporal data. [Method/process] Based on the knowledge graph theory, this paper takes the topic of“The launch of Tianhe core module”on Sina Weibo as an example and applies the technology of spatiotemporal big data mining to construct the endogenous graph. We further mined the spatiotemporal endogenous effect of public opin? ion evolution from time, space and attribute dimension. [Result/conclusion] Different attributes of public opinion sub? jects present multi-dimensional correlation patterns. The main body relationship of social network public opinion evo? lution shows obvious spatial differences based on different time series. The evolution of public opinion under this event conforms to the law of event evolution and has the characteristics of temporal and spatial differentiation. The research of this paper can improve the spatiotemporal data management ability of public opinion for supervision departments, ac? celerate the realization of intelligent public opinion guidance and control innovation mode. It can also provide guidance for the implementation of targeted control measures.