Methods for Developing Hierarchical Models

Research project, 1997-99, financed by the Ministry of Science and Technology of the Republic of Slovenia under grant 3411-97-22-9076.

Principal investigator: Marko Bohanec

Research Unit: Department of Knowledge Technologies, Jožef Stefan Institute, Ljubljana.

Summary: The project is aimed at the investigation and development of methods for an active support of designers in developing hierarchical and evaluation models. The proposed approach is based on (1) automatic and supervised development of models from learning examples based on function decomposition, and (2) dynamic transformation of hybrid qualitative and quantitative models. Expected contribution of the project is an integrated set of methods for supporting the development process at different levels that range from an active support in manual development of models to automatic development based on machine learning from examples.

Keywords: machine learning, constructive induction, structured induction, hierarchical models, multi-attribute decision making, function decomposition



This project was discontinued at the end of 1998, after two completed years of research. The termination was due to the reorganization of research projects in Slovenia, which required their transition and incorporation into "programs".