With the development of technology of digital and intelligent transformation, the management science has entered a new era. The information management, as a research area closest to information sciences and technologies in management science, has completed a new upgrade with the support of complex mathematical models, the upgraded version of the technology is more difficult and the model complexity is higher. This article will review and summarize the social network public opinion management in the upgraded version of information management, selecting three representative frontier directions: social network public opinion tracing, AIGC (artificial intelligence generated content) army identification, and early detection and trend prediction of public opinion. In the era of rapid development of information technology such as social networks and intelligent large language models, the characteristics of more sources of online public opinion information, unpredictable information quality, faster diffusion speed, and wider diffusion range have led to a sudden increase in the difficulty of managing online public opinion. This article will unfold from two main branches of social network public opinion management, "based on social network structure" and "based on social network content". The former will summarize advanced technologies based on public opinion tasks, while the latter will first decompose "content" based on differences in task content, and then summarize up-to-date technologies separately. Based on the social network structure, this article divides public opinion tracing and dissemination into four representative technologies aimed at different tasks of public opinion tracing and dissemination, and summarizes them: social network topology analysis and enhancement technology based on topology reinforcement learning for public opinion tracing, social network topology analysis and enhancement technology based on meta path graph representation learning for public opinion propagation path mining, social network topology analysis and enhancement technology based on capsule graph. Based on the direct analysis method of content and the direct analysis method of content, the "social network based content" is divided into "direct detection model of social network public opinion content information based on deep learning: AIGC detection model, rumor detection, and water army identification " and " indirect detection model of social network public opinion content information based on knowledge graph embedding: early public opinion discovery and trend prediction", respectively. The former takes the impact of AIGC, rumor and water army on public opinion management in social network as the basic point, which is divided into: AIGC detection model and experimental analysis for multi-modal public opinion information in social network, rumor detection method in social network, and water army identification technology in social network. The latter is based on the knowledge graph embedding model for early detection and trend prediction of public opinion that is decomposed into: public opinion analysis models based on static knowledge graph, public opinion analysis methods based on dynamic knowledge graph, public opinion analysis technologies based on knowledge hypergraph. The overall narrative logic of this article is 1 (structure) + 2 (content) = 3 (application issues of public opinion management), and below it, each problem is unfolded in order of hierarchy. Then, relevant representative work is reviewed, and finally refined and summarized.