Abstract
This paper proposes a new fuzzy integral model which can treat interactions among arbitrary macro-attributes as an effective modeling method. Since a partition of the set of all attributes defines macro-attributes, this fuzzy integral is named partition type fuzzy integral. An attainment degree to each subset of a partition (i.e., macroattribute) is evaluated by a vector, so the partition type fuzzy integral is formulated by utilizing the multilinear fuzzy integral of vector valued functions. This method evaluates the interactions of the attributes each of which belongs to a distinct macro-attributes, which means that the evaluation process can be simply explained by fewer parameters. We verify these features by applying this method to a subjective evaluation problem.
