| Abstract: | Problem solving is a style of thinking, which transforms a given problem to the goal state through a so-called PS (problem solving) path. Different from the traditional GPS (General Problem Solver) approach, the focus in this paper is placed on how to judge the performance of PS paths, that is, the evaluation problem of problem solving.A series of PS paths point from the given source problem to the destination goal, then form a PSN (PS Network). This paper proposes an elaborated CPSN (Coordinate Problem Solving Network) as the evaluation model of problem solving. In CPSN, each problem is assigned a unique coordinate and then each PS path can have an evaluation vector. Several examples show such arrangement can give more insight to PS paths.Furthermore, an incremental learning algorithm is developed for the update of CPSN. When a new PS path is obtained, it is not necessary to recalculate the whole CPSN. Examples show such algorithm provides a more efficient way in finding new PS paths. |
