MINES is a novel GraIL-based framework that enhances inductive relation reasoning over neighbor-enhanced subgraphs by introducing a Message Intercommunication mechanism. The framework addresses two key limitations of existing GraIL-based models: insufficient message communication and insufficient neighborhood information collection. To improve message communication, MINES introduces bi-directional information interactions between connected entities by inserting an undirected/bi-directed GCN layer between unidirectional RGCN layers. Additionally, the framework extends the neighborhood area beyond the enclosing subgraph to enhance information collection for inductive relation reasoning. Extensive experiments on twelve inductive datasets demonstrate that MINES significantly outperforms other models in terms of performance, effectiveness, and transferability. MINES achieves a 4.01% improvement in AUC-PR and a 2.78% improvement in Hit@10 across datasets. The framework also shows a lightweight design, reducing parameter count by about 1500 for each dataset compared to the prototype GraIL. Ablation studies confirm that the message intercommunication mechanism and neighbor-enhanced subgraph extraction contribute to improved performance. Transfer analysis on TACT and CoMPILE shows that MINES can be effectively extended to other GraIL-based models, demonstrating its scalability and generalizability. The framework's sequential intercommunication mechanism is shown to be more effective than parallel frameworks, with the RGR variant outperforming others in terms of AUC-PR and Hit@10. Overall, MINES provides a more expressive and discriminative approach to inductive relation reasoning in knowledge graphs.MINES is a novel GraIL-based framework that enhances inductive relation reasoning over neighbor-enhanced subgraphs by introducing a Message Intercommunication mechanism. The framework addresses two key limitations of existing GraIL-based models: insufficient message communication and insufficient neighborhood information collection. To improve message communication, MINES introduces bi-directional information interactions between connected entities by inserting an undirected/bi-directed GCN layer between unidirectional RGCN layers. Additionally, the framework extends the neighborhood area beyond the enclosing subgraph to enhance information collection for inductive relation reasoning. Extensive experiments on twelve inductive datasets demonstrate that MINES significantly outperforms other models in terms of performance, effectiveness, and transferability. MINES achieves a 4.01% improvement in AUC-PR and a 2.78% improvement in Hit@10 across datasets. The framework also shows a lightweight design, reducing parameter count by about 1500 for each dataset compared to the prototype GraIL. Ablation studies confirm that the message intercommunication mechanism and neighbor-enhanced subgraph extraction contribute to improved performance. Transfer analysis on TACT and CoMPILE shows that MINES can be effectively extended to other GraIL-based models, demonstrating its scalability and generalizability. The framework's sequential intercommunication mechanism is shown to be more effective than parallel frameworks, with the RGR variant outperforming others in terms of AUC-PR and Hit@10. Overall, MINES provides a more expressive and discriminative approach to inductive relation reasoning in knowledge graphs.