The European Journal on Artificial Intelligence

The interaction of the atmosphere and the ocean has a profound effect on climate, while the uptake by the oceans of a major fraction of atmospheric CO₂ has a moderating influence. By improving accuracy in the quantification of the ocean's CO₂ budget, a more precise estimation can be made of the terrestrial fraction of global CO₂ budget and its subsequent effect on climate change. An instance-based reasoning system, which incorporates a novel neural-based clustering and retrieval method has been developed for estimating the CO₂ budget. This paper reviews the problem of measuring the ocean's CO₂ budget, presents the model developed to resolve it and presents the obtained results.

SIADEX is a complex framework that integrates several AI techniques able to design fighting plans against forest fires. It is based on four main components, a web server, that centralizes all the flow of information between the system and the user, the ontology server, that is the cornerstone of the architecture as the basis for knowledge sharing and exchange between all the components, and the planning and monitoring servers that are offered as intelligent services through the web server. This perspective also allow to view SIADEX as a collaborative working environment where it provides two basic functionalities: the intelligent services offered to the user (the ontology, planning and monitoring modules) and a middleware level that interfaces these back-end services to the front-end software of the user (a web browser) achieving several valuable goals like transparent access of the user to a distributed architecture, an ubiquitous access to the services that allow the mobility of the user and his/her independence of the access device.

For the last years, artificial intelligence (AI) approaches have become useful tools in environmental engineering. Here, one relevant application area is the optimization of wastewater treatment plants (WWTP). In this paper, we present several examples for real-time Control (RTC) tasks and decision support systems (DSS) for wastewater treatment (WWT), specifically based on case-based reasoning (CBR). Moreover, we present an approach for optimizing the prediction accuracy of these systems. The idea of this approach is to employ knowledge-intensive similarity measures instead of simple distance metrics. In order to facilitate the modeling of these measures resulting in lower deployment costs of the CBR systems, we propose a novel machine learning technique.

We describe a Decision Support System for plant disease management used by technicians of the Advisory Service of Trentino region, Italy, and by the researchers in disease management techniques. We present the Artificial Intelligence (AI) methods and techniques that have been exploited during the development of the system. In particular, we discuss the role of an Agent Oriented analysis of the application domain, for requirements elicitation and system design, where AI supported the developers of the systems. In addition, Machine Learning techniques have been used to develop decision procedures to support the domain stakeholders in planning and executing actions for managing a plant disease. We illustrate results of the Machine Learning techniques application with reference to a critical apple pest. Notice that in this case AI techniques are part of the system itself. We also describe the system architecture and the main user functions.

Air pollution control in urban regions is one of the main directions of research in the environmental sciences. For each region the pollution causes and pollution dispersion are different, depending on the industrial activity, on vehicles traffic, on domestic sources and so on, as well as on the geographical location, temperature of the air, speed and direction of the wind, and other weather factors. Several mathematical models are used for the description of the relationships between environmental protection and meteorological factors. An alternative approach to the mathematical models is a knowledge-based approach, that integrate multiple sources of knowledge in a knowledge base.

The paper describes a case study of knowledge modelling in an air pollution control decision support system that uses DIAGNOZA_MEDIU, a prototype expert system dedicated to air pollution analysis and control in urban regions. We have developed an ontology, AIR_POLLUTION, for the application domain. Several AI techniques were used in the knowledge modelling process. An artificial neural network provides predictive knowledge to the facts base, and a part of the rules from the rule base are extracted by using inductive learning.

Ecological theories often explain the behaviour of communities in terms of the underlying interactions that take place between the species that are part of the community. This closely relates to the idea of compositionality within Qualitative Reasoning, in which the model that simulates the behaviour of a larger system is automatically assembled from previously defined elementary units. A recurring question however concerns the viability of this approach, particularly the scalability of such small partial descriptions. In this article we present a fully implemented model and accompanying simulations of the Ants' Garden, a complex system consisting of multiple species with multiple interactions. The Qualitative Reasoning engine automatically assembles these models by reusing a previously created library with partial models of basic processes that govern population behaviour and interactions. The simulations show the typical behaviour of the Ants' Garden as currently known and described by experts, and support the idea that our previously developed library is adequate and scalable to simulate complex system behaviour.

Clustering techniques have a great importance in knowledge discovery because they can find out new groups or clusters of objects within databases. Thus, they are unsupervised learning methods, very useful when facing unknown, unlabelled and ill-structured databases, as environmental databases are. In this paper, different clustering algorithms are analyzed and compared. They are used on a real environmental data set in order to study their impact in characterizing states in this kind of domains. The comparison of the methods is undertaken using the system GESCONDA, which is a prototype of a data mining tool. Environmental data used in this paper are from a Catalan wastewater treatment plant and refers to different variables of the plant at different spatial points along 149 days.