Underwater acoustic communication is the most reliable way of underwater long-distance communication. However, due to the harsh underwater environments, the battery capacity of underwater acoustic sensor nodes is limited and difficult to charge. The rapid energy depletion of nodes with high utilization rate causes the energy hole problem, which reduces the lifetime and throughput of underwater acoustic networks (UWAN). In this paper, we propose a method of data importance classification and forwarding of data hoarding on nodes, which uses an autonomous underwater vehicle (AUV) equipped with a reconfigurable intelligent surface (RIS) system in the energy hole area. The proposed method addresses the issues of data transmission and data accumulation at nodes within energy holes. By utilizing the RIS system, data can be rapidly forwarded, and RIS-based chunking techniques are applied to handle data of different importance levels. This approach efficiently accomplishes targeted forwarding of accumulated data. In addition, this study utilizes genetic algorithms to address the chunking problem in the RIS system based on data importance, effectively tackling the challenges of large search space and lack of gradient information in RIS chunking optimization.