Datasets
When we have started to develop HINT,
our implementation of function decomposition, of a great
help where data sets that were constructed from multi-attribute
decision models originally developed using DEX methodology. In DEX,
the expert manually constructs concept hierarchy and defines
functions that relate these concepts with attributes. The formalism
used to describe the resulting model and its interpretation are
essentially the same as those derived by decomposition. This makes
models developed by DEX ideal benchmarks for the evaluation of
decomposition.
There are two data sets together with
their concept hierarchy available:
Note that both data sets (but not their
concept hierarchy) are also available on the UCI ML
repository.