[Purpose/significance] The visual analysis of the topic map of network rumor relation path in the big data- driven public opinion propagation of social network can play an important guiding and promoting role for the relevant public opinion management departments to effectively supervise and correctly guide the public opinion of social net? work.[Method/process] In this paper used the web crawler to obtain data on Internet rumor topics in the dissemination of public opinions in social networks, and based on the topic map of opinion leaders to construct the entity and user rela? tionships of Internet rumors, determined the relationship path analysis parameters of opinion leaders. Taking Sina Wei? bo topic of "chongqing bus's fall into the river·non-female driver's retrofitting" as an example. Neo4j, an open source knowledge graph tool, is used to construct the theme graph of data, and Cypher language is used to analyze the commu? nication efficiency, communication path and influence of key nodes of opinion leaders. [Result/conclusion] Data re? sults show that knowledge map of opinion leaders is a kind of directed network relation graph in Weibo network rumor propagation. Opinion leader nodes have a strong influence, and the communication of ordinary user nodes is extremely susceptible to the influence of opinion leader nodes. Opinion leader is the key figure in the propagation of rumors. At the same time, the knowledge map of social network public opinion presents a secondary communication trend. A single opinion leader node does not have the ability to control the spread of rumor topics. The barrier between opinion leader nodes and ordinary user nodes will slowly weaken to a certain extent with the continuous spread of network rumors and public opinions, so that social network public opinions will spread in a flatter and disorderly way.