![](bg3.png)
2.3 Search Engine
The metadata search engine in MDWeb provides
users various research modes, one of the most use-
ful is the ”multiple-criteria searches”, which allows
to compose a query based on four criterias :
• What: Allows the user to specify one or more
keywords.
• When: Allows the user to specify the period in
which the reference was created.
• Where: Allows the user to restrict the search to a
specific geographical location of data.
• Who: Allows the user to specify an administra-
tive entity (agency, institution, etc.) of data.
2.4 Summary
The MDWeb project focused on sharing and mutual-
izing existing data, we disposed, as presented in this
section, a component for managing metadata and a
search engine for localizing data based on these meta
informations. In short term, we will also extend the
same functionality for the treatments. The next im-
portant step for us, is to focus on the workflow com-
ponent, as introduced in the section 1.2.
3 OUR VISION OF A
WORKFLOW ENVIRONMENT
Our objective is to integrate the workflow environ-
ment in the sharing and mutualization platform. From
the business point of view (experimenters), workflow
usage corresponds to the three stages shown in figure
3):
Figure 3: Business point of view.
1. Definition: abstract definition of a process chain
corresponding to an experimentation (planning of
experiments),
2. Instantiation: more specific definition after
identifying the various elements of the chain
(data/processes),
3. Execution: customized execution (according to
strategies corresponding to the requirements).
It is from this experimental life cycle and inspired
by architectural styles proposed by OMG (OMG,
2006) that we have proposed the 3-level architecture
(cf. fig. 4):
Figure 4: A workflow environment.
1. The static level concerns the design phase and
consists of constructing business process models
(abstract) using a simple language defined by a
meta level. There exists several standards and
specifications for defining a process model. In
(Lin et al., 2009), we have analysed some of them
like : UML (Activity diagram) (OMG, 2001),
SPEM (OMG, 2005), BPMN (OMG, 0082). One
common point of these standards and specifica-
tions is that, they are all very comprehensive but
require substantial time to understand and to use,
in order to take full advantage of them. However,
it might be not so easy for scientists which are not
experts in this field.
2. The intermediate level represents an instantiation
and pre-control phase. Before going on to the
execution phase, the users should refine and cus-
tomize their experimentation based on the busi-
ness process model defined during the last step,
by determining and localizing the most suitable
sources of data, programs and services. After
that, we propose to include pre-control within this
phase, for guaranteeing the executability of the
generated process chain. The objective of this pre-
control is to verify the validity of the instanced
chain based on formal conformity rules. We then
continued our studies by analysing the different
scientific workflow environments like Kepler (Al-
tintas et al., 2006), Taverna-myGrid (Hull et al.,
2006), BioSide (BioSide, 2008), etc. Each project
declines an graphical interface for defining work-
flow (using a abstract syntax not necessarily ac-
cessible for users), and then, during the execu-
tion of a defined workflow, they handle the confor-
mity problem by using specific adaptations, either
manually or semi-automatically.
3. The dynamic level concerns the actual execu-
tion phase. This takes place according to various
strategies defined by both the experimenter and
the operational configurations.
ICEIS 2010 - 12th International Conference on Enterprise Information Systems
52