Abstract: Binary discernibility matrix-based reduction approaches are characterized by vivacity and operability，which make them widely applied to projects.However，in existing approaches，binary discernibility matrixes with large scale are usually produced when scanning data set，and are used for reduction.This leads to excessive consumption time.In this paper，the scale of binary discernibility matrix are greatly reduced by using absorption law on row and logic addition，and a reduction algorithm based on binary discernibility matrix is proposed.This algorithm is characterized by good operability and feasibility，and its time and space complexity are relatively reduced.In the developing project，the proposed algorithm has been used，which is used to acquire knowledge in detected data of liver function，and the result is satisfactory.