Applying domain knowledge to the discovery of operating situations in wastewater treatment plants

LINNEO⁺ is the implementation of a classification methodology for concept acquisition from non-classified examples in ill-structured domains. In order to achieve this aim, LINNEO⁺ uses incremental conceptual clustering and domain knowledge expressed by means of experts´ rules. The final target is to build knowledge bases or case libraries after an expert´s validation. The improvement of the results during classification process biasing using partial expert knowledge (classification rules) is presented. Some experiments are shown concerning the effect of different sets of expert rules on the process. The effect of the domain knowledge on the order of the examples is also studied. An application to the characterization of the operating situations of wastewater treatment plants is used to show the performance of the system. The stream of data used in this application is a collection of measurements of a real wastewater treatment plant located in Manresa (Spain)