KATE is a CBR and Induction system developed by Acknosoft. It supports a simplified view on the the CBR process, where focus is on retrieving cases that match the description of a new case, using a nearest neighbor algorithm. KATE handles complex data represented as structured objects, relations or general domain knowledge. But it can not explain similarities or explain away differences by other methods than with a similarity metric. The solution method must be reused ``as is'', or can be changed by the user, KATE does not offer possibilities for automatic adaption of the solution. The revise step is not possible to do within KATE, other than checking the casebase for counter-examples. The retain step consists of adding cases and modifying weights for features.
However, KATE is an efficient industrial implementation which offers possibilities to weigh features when computing similarity metrics, define a similarity matrix or to use a user-defined similarity metric. It is also easy to expand because all functions of KATE are available through DLL library functions.
The input for KATE is a domain file which contains descriptions of features and feature values in the CASUAL language, which was developed in the Esprit INRECA project [7].
The version of KATE used here is 6.0 for Windows 3.11. KATE and other tools for CBR are reviewed in [6].