摘要 |
PURPOSE:To automatically or semiautomatically learn the attribute in hierarchy of conception that should have been defined by a human being before. CONSTITUTION:A learning part 1 receives an input from an input part 4 by a question to show whether a certain concept has a certain attribute and refers to a storage part 3 of the hierarechy conception to calculate the information amount that is obtained when it is known whether another concept in the hierarchy of conception has the precedent attribute. Thus concept of the largest information amount is found and shown at a display part 5. Then, the part 1 receives the input from the part 4 to show whether the concept of the largest information amount has the relevant attribute. These operations are repeated by several times so that the attribute of the concept is learnt in the hierarchy of conception. Then, a learning part 2 refers to the hierarchy of conception and decides by generalization whether another concept in the hierarchy conception has the relevant attribute based on an MLD standard, etc. Thus the part 2 learns the attribute in hierarchy of conception. |