20 This arTicle has been peer-reviewed. Computing in SCienCe & engineering
S e c t i o n t i t l e
Problem-Solving Methods for
Understanding Process Executions
Jose Manuel Gómez-Pérez
iSOCO
Oscar Corcho
Universidad Politécnica de Madrid and University of Manchester
Problem-solving methods are high-level, domain-independent, reusable knowledge
templates that support the development of knowledge-intensive applications. The authors
show how to use them to bolster subject-matter experts’ understanding of process execution
by implementing such methods into the Knowledge-Oriented Provenance Environment.
I
n the context of scientific data for compu-
tation-intensive disciplines such as physics,
biology, and astronomy, provenance focus-
es on describing and understanding where
and how data is produced, the actors involved in
its production, and the processes applied before it
arrived in the collection from which it’s now ac-
cessed. In a typical discovery task, for example,
scientists integrate data from various sources, fil-
ter the combined data according to some criteria,
and then annotate it with information about the
relationships they’ve just discovered. All the tasks
applied in this process contribute to that data
product’s provenance record.
However, having all this information recorded
together with the data product isn’t enough—given
the large amount of information, the provenance
record requires an abstraction process before any-
one can use it. Think of provenance information
as a pyramid with four levels from the bottom up:
data, organization, process, and knowledge.1 Al-
though most current provenance systems focus
on the first three levels by providing means for
recording and querying process documentation,
other efforts approach the provenance problem
from a semantic perspective in an attempt to
tackle the knowledge level. These systems use
domain ontologies in Semantic Web languages
such as RDFS (www.w3.org/TR/rdf-schema) and
OWL (www.w3.org/2004/OWL), which establish
well-defined associations between the resources
used during process documentation and the do-
main. This lets users build semantic provenance
metamodels with the terminology necessary for
meaningfully expressing provenance entities and
the relationships between them.
But regardless of the approach taken for prov-
enance gathering and representation, the docu-
mentation of a process’s execution generates large
quantities of heavily linked and annotated prove-
nance data. As the size and complexity of processes
increase, process documentation can become hard
to assimilate and eventually unmanageable. Fur-
thermore, the main beneficiaries of provenance
information are subject-matter experts (SMEs)
who don’t necessarily have a strong background
in computer science or, more specifically, prove-
nance. An additional semantic layer with a higher
level of abstraction could help address this gap.
As much as possible, our goal is to support
1521-9615/08/$25.00 © 2008 ieee
Copublished by the IEEE CS and the AIP