Analogical reasoning for 3-D prediction in proteins

Biological entities present a complexity level that should be correctly managed in Artificial Intelligence environments. In the paper, three-dimensional structures prediction is presented on a calculability point of view. One has to describe and access proteins in a proper way. This is done by the means of specialized structured editors, following direct manipulation principles. Objects classes and properties are selected by experts in order to constitute the learning focus. They lead to regularities determination on sets of biological entities. The end-user refines his knowledge by objecting to the results of the system. Analogous reasoning strategies are followed, which allow to closely control the validity and pertinency of the acquired knowledge. An environment prototype is presented in which learning should be considered as a basic tool as well as database access or editing