Four different demos, based in four Test Cases defined in deliverable D.7.1 (Experimentation and Evaluation Plan), are presented in this section:
The Test Case this demo is based on, is a variant of the Use Case EOSelect, described in section Scenarios.
In our Test Case, a requestor actor asks for a valid image for detecting oil spills. In this first Test, we considered the EORequestor-Crisis actor, whose main concern is to get a valid image as soon as possible. Price, format or other characteristics are of secondary importance.
The expected output for this Use Case was the selection of the fastest image provider among all the valid and available image providers at the best price. This price is negotiated using the minimal concession protocol (MCP), as defined in deliverable D.4.1 (Towards argumentation-based contract negotiation).
 |
In the Test Case used for this demo, based also in the Use Case EOSelect, the EORequestor-Research actor is interested in images of high quality, with resolution within given boundaries, and with price being a strong point in its decision. Cheap images should be preferred to expensive ones. A specific price range is specified by the EORequestor-Research actor.
The ideal output for this Test Case is the selection of an image with optimal relation between price and quality, according to the EORequestor-Research preferences (resolution boundaries and price interval). Assume however that no provider exists matching these preferences. The output is then a sub-optimal but sufficiently close match, that the user is presented with nonetheless. This sub-optimal solution is negotiated using the minimal concession protocol with rewards (MCPR) (in deliverable D.4.1). This sub-optimal solution is the result of negotiation of workflow (in the sense of image quality) and contract (price).
|
The Test Case this demo is based on, is a variant of the Use Case EOComp, described in section Scenarios.
A requestor actor will ask for a fire detection product that has to be created from existing Image Providers and Image Processors.
This test is analogous to the EO Selection MCP, in that we will consider a crisis situation where the main concern of the EORequestor-Crisis agent is to get a valid product as soon as possible. Price, format or other characteristics are considered as secondary. However, the expected output for this Use Case is a combination of services (an image, a clipping, and a fire detection processing).
The combination is obtained from the services that give the required fire detection product as quickly as possible. The prices of all services in the combination are negotiated using the minimal concession protocol (MCP) (in deliverable D.4.1). In addition to the EORequestor-Crisis agent, this test also involves ImageProvider agents and ImageProcessing agents (including clipping agents and fire detection agents).
|
For the fourth demo, another variant of the Use Case EOComp is used.
The EORequestor-Research actor is interested in final products with higher quality or closer fit to its preferences. Price is a strong point in its decision and cheap image providers and image processors will be preferred to expensive ones.
The expected output for this Test Case is the selection of a final "fire detection" product with optimal relation between price and quality, according with EORequestor-Research preferences, if one exists, or the best quasi-match otherwise
|
