 |
ESAT: Selecting Assisting Technologies for Handicaped People#
ESAT ("Expertensystem für Assistierende Technologien" in German) is an expert system designed to determine an appropriate set of human-computer interaction devices for handicapped people. In the application scenario a detailed profile of the physical capabilities (e.g., visual or motorical abilities) for a person is entered into the system. The knowledge base derives a set of input and output devices, that together provide optimal computer interaction for that specific person. In advance, the underlying domain knowledge has been elaborated by a comprehensive study in 2008. The actual implementation of a corresponding executable knowledge base using KnowWE has started in spring 2011.
Currently, the ESAT knowledge base has been completed and the system will be launched for a testing phase at the project’s initiator (http://www.vo-fab.at/
). The knowledge base has been implemented by mainly one knowledge engineer using KnowWE. For knowledge representation production rules are used. In total the ESAT knowledge base currently contains 654 rules distributed on 74 wiki articles. Also in this single-user context the possibility of free structuring allows for reasonable and clear distribution of the knowledge. The terminology is defined on different wiki articles dealing with vision, hearing, motoric and haptic abilities and general skills (e.g., braille) respectively. The about 50 different types of input and output devices (e.g., various kinds of keyboards, sensors, displays) are each described in distinct wiki articles also containing the rules relevant for the derivation of the particular device. Five heuristics have been established within a theoretical study, describing solutions for major categories of handicaps. These are implemented on distinct wiki articles forming the core of the derivation knowledge.
The testing framework for continuous integration discussed in Section 3.3 is extensively used to guarantee the save development process by uncovering undesired side-effects of modifications including at least one sequential test case for each device and heuristic. More details about the project are given by Kreutzer.
