Abstract: The requirement of deep language understanding is more and more important in the fields of information auto-processing and information security. To solve the difficulty of computer understanding Chinese sentential semantics, this paper proposes a computable and extensible Chinese sentential semantic mode (BFS-CSM), which is based on the theory of modern Chinese semantics. In addition, a reversed question based sentential semantic structure method is put forward to verify the correctness of proposed BFS-CSM. Experimental results show that the precision of the verification method is 92.07% and the rationality of proposed mode is confirmed. This research provides a new way of thinking for Chinese natural language processing and has important applied value.