Investigation on AQ11，ID3 and the Principle of Discernibility Matrix

The principle of discernibility matrix serves as a tool to discuss andanalyze two algorithms of traditional inductive machine learning, AQ11and ID3. The results are: (1) AQ11 and its family can be completelyspecified by the principle of discernibility matrix; (2) ID3 can bepartly, but not naturally, specified by the principle of discernibilitymatrix; and (3) The principle of discernibility matrix is employed toanalyze Cendrowska sample set, and it shows the weaknesses of knowledgerepresentation style of decision tree in theory.