Towards a formal framework for the specification hybrid fuzzy modeling

This architecture starts when a previously existing software system ends. This software tool offers the needed functionality to work with general purpose inductive learning. Now, new functionalities, those needed to cope with hybrid fuzzy modeling, are being added to it. In order to reach our main goal, the following four ones are the most important ideas that we will consider. The first one is that we see data driven fuzzy modeling as a set of successive transformations done to data, starting from the raw source data and finishing at the desired fuzzy inference system. The second one is the modelization of the "fuzzy modeling task" concept in terms of domains, codomains, preconditions, postconditions inside a concrete domain of discourse. The third one is the definition of different schemes for tasks composition to form higher level tasks in a transparent manner. Finally, the fourth one is the concretion of the methodological steps to follow on the integration into this new framework of base techniques to be used as building blocks to compound higher level tasks. All these ideas are finally illustrated with an example.