Aggregation of Fuzzy Conformances

Retrieving the most suitable items and sorting them downwards from the best face many challenges. The conformance measures are able to efficiently calculate similarities between the desired value and values of considered items’ attribute regardless of different data types. These measures should be suitably aggregated, because the users usually provide different preferences among atomic conformances and therefore various aggregation functions should be considered. In this paper, we examine conjunctive functions (including non t-norms) as well as averaging and hybrid ones. In the hybrid aggregation, uninorms and ordinal sums of conjunctive and disjunctive functions have shown their perspectives in aggregating conformance measures. Diverse tasks require functions of desired behaviour and properly assigned weights or parameters. Thus, the perspectives for merging aggregation functions with the machine learning to the mutual benefits are outlined.