Hypothesis Selection using Domain Theory in Rule Abductive Support for Skills

Abstract

In skill science framework, We have studied for supporting of performer - mainly cello player - with employing logical computation systems. In one hand, we suggested to derive hypotheses rep- resenting possible ways to given certain performance tasks using an Abductive Logic Programming (ALP) system. On the other hand, we suggested method called rule abduction, which find possible rules explaining observed 'knacks' as reasonable performing method. There remain many things those must be solved. For example, we should show some directions to select better hypotheses from derived by ALP or rule abduction. In this article, we deal with further problems. When novel rules are abduced by rule abduction, some of them would have undefined nodes those should be given some adequate meaning. We build 2 databases for cello playing domain which can pro- vide information for aiding decision of meaning of such nodes by the performer. Using the same databases, we try selecting hypotheses derived by our existing system.