Adjustment of identified fuzzy measures

A fuzzy measures and fuzzy integrals model is often applied to a human evaluation model. For the construction of the human evaluation model it is necessary to identify fuzzy measures by the use of input-output data. In this paper an algorithm for the adjustment of the identified fuzzy measures is discussed. The algorithm considers super-additivity or sub-additivity of fuzzy measures. A user interface is designed to help adjust fuzzy measures so that the errors between the data and the outputs of the evaluation model are made uniform. The contradictory data can be found out by the present algorithm and the user interface. In order to construct a better human evaluation model the contradictory data are fed back to human subjects for a re-evaluation. The fuzzy measures are identified again using the re-evaluation data. The effectiveness of the adjustment of fuzzy measures is shown by evaluation examples about apartments. In order to confirm whether the evaluation model fits human evaluation or not, errors between the model outputs and questionnaire, and the analysis results of obtained fuzzy measures are considered.