Abstract
We proposed the correlational analogy method to assist architectural design process in previous papers. By this method, an associational relationship between two specific items A and B can be found out from a plan statistically and the discovered relation (A, B) means "B should be specified if A is specified and A should be if B is at the same time". However, we observe frequently the case that B is not always specified when A is specified although A must be specified under B. Also, there are 15 types of logical relation between two specific items. The new method is based on the observation of coincidental occurrences of every couple of specifications in the same plan as like as the correlational analogy method. After making plans by a designer, that is, after observation of frequency data of every couple of specific items from a set of plans, we consider that the observed data were obtained from a combination of the above 15 relation types. The highest possibility combination can be calculated with the frequency data. The relation type that has the largest number of occurrences in the most possibility combination is estimated as the most likely relation. The above discovered relation is a kind of personal knowledge and we can use the knowledge for assisting the same designer in design process. For example, if we discovered a relation (A>B) which means "B should be specified under specification A" from a set of plans designed by a designer, then a CAD system associates "B" when the designer specifies only "A".
