A problem solving reasoning is characterized by a set of actions that must be executed in order to achieve the problem's solution. Given a problem, one of the three following cases can appear: either its solution is known, or a solution of an analogous problem is known or, finally, a solution of a more generic/specific problem is known. So, a reasoning similarity evaluation method is desirable because this allows to reuse, to adapt or to modify known reasonings and, consequently, makes simpler the search of a problem's solution.

         One of the techniques often used to solve a problem is to decompose it in subproblems until to reach elementary problems whose solution is easy or known. These successives decompositions can be represented by a graph where the vertices represent the tasks (a task represents a problem and its solving process) and the edges symbolize the decomposition and the specialisation links between tasks. Based on this reasoning representation model, task languages are developed. These languages can be used for creating problem solving environments which goal is to completely automate the problem resolution proccess. The input/output data as well as the solving strategy (actions to do, their conditions of execution and data flow) are the essential attributes of a task. Given that the reasonings are represented by task graphs, to measure the similarity between two reasonings means to evaluate the isomorphism between two task graphs.

   [ My first page] [ Curriculum Vitae] [ Projet's Page ][ Penduick-FM]