A Logical Splitting Strategy for the Belief Structure of Agents

We prove that for any finite deduction structure there exists a unique concise-widest chain-preserved split. Based on this result, we propose a logical splitting strategy which enables an agent to split its belief structure such that all the original inference chains can be preserved. The significance of such logical splitting at least is four-fold: (1) It can be used by an agent to separate its concerns appropriately, or even create smaller and smarter clones which could save time and efforts in their deliberation computing; (2) It will enable an agent to adjust its concerns dynamically based on its recognition of the current situation, which may further enable the agent to make rapid and rational decisions when situation shifting occurs; (3) In periodic team synchronization (PTS) domains where agents can periodically synchronize their behaviors with no restriction on communication, it can be employed as a critical process for maintaining the shared mental models among a team of agents; and (4) It is helpful in alleviating the complexity of reasoning about relevant information, and in better anticipating both explicit and implicit information needs based on the recognition of the current situation.