Keywords

environmental decision support systems, knowledge based system, wastewater treatment plant, hybrid systems

Start Date

1-7-2012 12:00 AM

Abstract

Given the complexity of Waste-Water Treatment Plants, both from the environmental, legal and economic point of view, Environmental Decision Support Systems (E-DSS) are getting wider adoption to monitor and manage the plants in real time. From a cognitive perspective, the knowledge required by an (E)DSS may be encoded in different forms. In this paper, we argue that the operational domain and its most relevant concepts should be defined in a proper ontology, providing a vocabulary to encode inferential or operational knowledge in the form of decision-making rules. The rules process information extracted from data, acquired through sensors and possibly processed using predictive or analytic models. Eventually, the rules themselves and the actions they recommend, can be orchestrated as business processes, using workflow models. Moreover, we argue that standard formats should be used to facilitate the formalization and exchange of knowledge between different systems, including OWL(2) (ontologies), RIF/RuleML (rules), BPMN(2) (workflows) and PMML (predictive models). Finally, we present a use case modelling a periodic plant monitoring routine which is necessary to check that the plant emissions are compliant with the national legislation. The system, implemented using the open-source Knowledge Integration Platform Drools, exploits a hybrid knowledge base but relies on a unified data model and execution environment.

Share

COinS
 
Jul 1st, 12:00 AM

Ontologies, Rules, Workflow and Predictive Models: Knowledge Assets for an EDSS

Given the complexity of Waste-Water Treatment Plants, both from the environmental, legal and economic point of view, Environmental Decision Support Systems (E-DSS) are getting wider adoption to monitor and manage the plants in real time. From a cognitive perspective, the knowledge required by an (E)DSS may be encoded in different forms. In this paper, we argue that the operational domain and its most relevant concepts should be defined in a proper ontology, providing a vocabulary to encode inferential or operational knowledge in the form of decision-making rules. The rules process information extracted from data, acquired through sensors and possibly processed using predictive or analytic models. Eventually, the rules themselves and the actions they recommend, can be orchestrated as business processes, using workflow models. Moreover, we argue that standard formats should be used to facilitate the formalization and exchange of knowledge between different systems, including OWL(2) (ontologies), RIF/RuleML (rules), BPMN(2) (workflows) and PMML (predictive models). Finally, we present a use case modelling a periodic plant monitoring routine which is necessary to check that the plant emissions are compliant with the national legislation. The system, implemented using the open-source Knowledge Integration Platform Drools, exploits a hybrid knowledge base but relies on a unified data model and execution environment.