|
Science is about understanding! Although Data Mining (DM) has been extremely useful
in a large number of domains where data analysis is necessary, most of the concerns
of the DM users has been the development of highly accurate models according to some
pre-specified metrics. Those concerns are completely justified and enough for a wide
range applications like predicting stock exchange market, for example. However, in
Scientific applications like the Live Sciences or Chemistry a good performance according
to the evaluation metric is most often not enough and an explanation for the phenomena
that produced the data is required. DM algorithms that produce Symbolic models, or
approaches that extract useful knowledge from black-box models are very useful tools to
propose explanations that could help experts to better understand the domain.
This workshop addresses such concerns. The workshop will be concerned with
the exchange of experience among researchers and provide updated knowledge concerning
the extraction of useful domain knowledge either by directly using symbolic DM algorithms
or post-processing non-symbolic ones.
The topics of interest include (but are not restricted to) the following ones:
|