Our goal is to develop:

• Specific 'knowledge access' interfaces - thus assuming that it shall be feasible to standardize knowledge formalization. De-facto standards to represent wide types of knowledge are now available, from rule and complex ontology to neural networks.
• 'Knowledge services', categorized as general purpose to grant the effective handling of the knowledge.
• 'Domain Knowledge Enabled Services', which are those applications specific constructed with all the general purpose available services (data, information and knowledge ones).
• 'Applications: knowledge exploitation and formalization'. Resulting from the combination of knowledge services as well as applications, which allow the system to be 'instructed', support (supervised/unsupervised) learning or - in general - bring knowledge into the system.
• 'HMI knowledge formalization and application'.

Figure 1 depicts the architecture of the EO-KES system, which most relevant aspects are:

• The EO-KES system uses a set of EO Data archives available via regular Internet connections
• The EO-KES will be a repository of knowledge about EO domain
• EO Data Processing can be executed, by accessing data from the archive, and information can be generated using the domain knowledge captured in the EO-KES.
• Information Mining ideas can easily be fitted into this architecture
• Web clients (EO-KES expert or the EO-KES) communicate with the EO-KES Server via a standard Web Server. Web clients are selected to provide a simple, scalable and straightforward way of connecting to EO-KES.
• EO-KES is by definition as well as by implementation an aggregation of distributed, self-organized (Knowledge Enabled) services.

EO applications include widely known, potential, emerging and operational services. In this project we will take into account, as existing and emerging EO applications at least those related to “standard” applications (e.g.: oil spill detection, ship detection, crop monitoring, flooding detection, terrain movement and subsidence detection, etc.), and forthcoming applications (such as those considered in the DUP, EOMD and GMES programs, as they become available).

The complexity of this project deserves a careful analysis about the approach during its analysis, design, and prototyping phases. The first point of reference is domain. Domain is – obviously – Earth Observation; nevertheless, a number of ‘concretions/focalizations’ are introduced. The following figure 2 illustrates this progressive focalization and represents the origin of the ‘scale’ factor, which makes the project ‘affordable’. Domain is a very relevant concept in this proposal, since it defines the boundaries of the ‘ontology’ (the knowledge of the experts has to be formalized).

The project is oriented in accordance with the limits of the domain. In the first view (until – latest - SRR), we consider a relevant contribution of EO domain experts directly supported by Knowledge Engineers. It is our comprehension that the very essential goal of this first ‘phase’ is to structure the problem: i.e., the basis of the domain ontology. In order to be in the position to match this goal, the proposal includes EO Specialists and Knowledge Engineers.

The second view (until PDR) refers to EO-KES. EO-KES represents a sub-set of functionalities, particularly those related to Knowledge Systems (Knowledge Enables Services). At that point, major contributions shall be expected from Knowledge and Software Engineers: How to gather, discover, formalize and use ‘that’ particular knowledge? And how to implement all these?

The third view (until FP) is centred in the EO-KES Prototype. The EO-KES Prototype is – once again – a subset, a representative subset, of the functionalities designed for the EO-KES. Implementation, testing and validation are the core issues.

• Earth-Observation
  
• Knowledge Based Systems
  
• Knowledge Engineering
  
• Machine Learning
  
• Computer Vision
  

• GTD - Ingenieria de Sistemas y Software Industrial, S.A.,Barcelona, Spain.
  
• Starlab, Barcelona, Spain.
  
• Tor Vergata Università, Rome, Italy.
  

• A. Pereira, R. A. Ribeiro, P. A. Sousa, M. Pantoquilho, P. Bravo, A. Falcão, S. D'Élia. An Ontology to Support Knowledge Enabled Services on Earth Observation. Proceedings of the ESA-EUSC 2004: Theory and Applications of Knowledge Driven Image Information Mining, with focus on Earth Observation Madrid, Spain, March 17-18, 2004 (to be published in ESA special publication number 533)