[Objective] Mining activity has posed a great impact on the environmental background of the areas. To meet the challenge, the near-natural recovery method might be the only feasible way. Soil microorganisms not only affect the material cycling and transformation of soil substances, but also play an important role in the restoring ecological systems. So to improve resilience of the mining ecological system, it is essential to clarify how soil microbial communities adapt to the environment and evolve therein; how do different microbial groups interacted with each other; and which of their functional groups are more suitable for the arid and barren environments? Moreover, knowledge about interactions between soil microbial communities and their responses to coal mining subsidence are crucial to ecological restoration and resilience of the ecosystems in the semiarid and/or damaged mining areas. However, so far little has been done in this aspect. [Method] To explore the structures and compositions of soil microbial communities, and mechanisms of reciprocal feedback between the dominant microbial groups and the environment in the subsidence areas of the Dongping (DP) and Dalita (DLT) Coal Mines, the methods of high-throughput sequencing and molecular ecological network were adopted. [Result] Results showed that the subsidence environment posed a significant impact on the diversity, structure and distribution of the soil microbial communities. Actinobacteria remained to be the dominant phylum in the DP and DLT Coal Mines, whereas DP had two more dominant phyla than DLT did, namely Nitrospira and Cyanobacteria. In the soil of DP, RB41, Solirubrobacter, Roseiflexus, Gaiella and Lysobacter were the genera > 1% in relative abundance, while in the soil of DLT, Nocardioides, RB41, Solirubrobacte and Roseiflexus were. The molecular ecological networks in the soil of the DP Coal Mine interacted more complicatly than those in the soil of the DLT Coal Mine, with more network nodes and connections. The soil microbial groups in the soil of the DP Coal Mine belonged to the phyla of Actinobacteria, Chloroflexi, and Thaumarchaeota, whereas the key species in the soil of the DLT Coal Mine were of the phyla of Proteobacteria and Actinobacteria. Moreover, significant relationships were observed between network structure and soil properties in both mining areas. For instance, among the DP networks, Modules 1, 2 and 3 were significantly and positively related to the soil EC, while module 2 was to the soil water content; Module 6 was significantly and positively related to soil available phosphorus, and Module 1 was to nitrate-nitrogen content; Modules 6 and 7 were significantly and positively related to soil dehydrogenase activity, while Modules 2 and 5 were to activity of soil urease. And among the DLT networks, Module 1 was significantly and positively related to pH and soil temperature, while Module 2 was to soil temperature only; Modules 1 and 2 were significantly and negatively related to soil water content and clay percent, whereas Module 5 was to nitrate-nitrogen content, but reversely.[Conclusion] In order to adapt to the nutritional infertility of the poor soil in the subsided mining areas, soil microbial communities tended to interact with each other. Holophagae and Aquabacterium were found to be the dominant species, and/or applicable to restoration of the ecological systems in the mining areas in the future.