![](bg5.png)
(ESB). ESB essentially acts as a backbone providing
high-level and reliable message-exchange services,
and transparently handles mediation of endpoint
heterogeneities and physical details during
component communication. Every time an EE is
produced by some functional component located in
the OS layer, the associated EAI snippet reactively
sends a message throughout the ESB. The message
encodes ad-hoc information, including the kind of
event occurred, the person or system that has
originated the event itself, the work context (e.g., the
actual project) of the event. Throughout the ESB, the
message is forwarded to any other EAI snippet that
subscribed that kind of EE. In turn, the latter EAI
snippet will consequently update its data repository
at the OS layer.
In addition to providing support for the
coordination of operational systems and data
integration tasks, in our proposed knowledge-based
framework EEs are also treated and traced as basic
units of work for the ex-post analysis of LSCPs,
being this analysis based on process mining and
OnLine Analytical Processing (OLAP) techniques.
The latter mining functionalities are supported by
the Process Discovery (PD) module, located at the
D&WS layer, and the OLAP module, still located at
the D&WS layer of the architecture, respectively.
Furthermore, D&WS layer also supports advanced
knowledge browsing, visualization, analysis, and
querying services, which are definitely able of
enabling effective decision making and collaborative
work tasks based on data and knowledge stored and
elaborated in the D&AI and KM&D layers of the
architecture. The latter functionalities are fulfilled by
the Knowledge Browsing and Querying (KB&Q)
module, located at the D&WS layer.
In our proposed knowledge-based framework for
LSCPs, data and knowledge are thus distributed
across the D&AI and KM&D layers of the
architecture, in order to augment the synergy among
all the components of the framework. More
specifically, for what regards data, the Enterprise
Data Warehouse (EDW), located at the D&AI layer,
contains snapshots of relevant enterprise data and
historical EE logs, which are represented in an
integrated and consolidated way according to the
EKM. In order to populate the EDW, the Enterprise
Data Loader (EDL) module, still located at the
D&AI layer, continuously elaborates all messages
exchanged throughout the ESB with the goal of
extracting data and storing them in the underlying
warehouse. To this end, canonical Extraction-
Transformation-Loading (ETL) primitives can be
advocated.
For what instead regards knowledge, the
Knowledge Manager (KM) module, located at the
KM&D layer, is in charge of maintaining a series of
interrelated ontologies (which are modeled
according to the ontology-based framework we
describe in Sect. 4) within an appropriate
Ontological Knowledge Base (OKB), still located at
the KM&D layer. Beside constituting a semantic
background that turns to be very useful for data
integration purposes, ontologies stored in the OKB
also enable a meaningful semantic annotation of
data stored in the EDW, while also nicely supporting
a semantic-aware access to them. More precisely,
the system allows to introduce meaningfully
interrelated ontologies for the EEs occurring in the
target (virtual) organization, which also capture
organizational concepts/entities associated with the
EEs themselves. Several studies have already
evidenced that collaborative processes clearly
benefit from the introduction of knowledge
management approaches and strategies. This
because the latter can effectively and successfully
support the management of knowledge that is
created, stored, shared and delivered along the
execution of collaborative processes. Therefore,
making use of a suitable ontology-based framework
within the knowledge-based framework we propose
(particularly, at the KM&D layer) is completely
reasonable, while it embeds several points of
research innovation. From a technical point of view,
the proposed reference architecture relies on the
ontology-based modeling framework presented in
(Gualtieri & Ruffolo, 2005), which both provides a
semantic infrastructure for the management of
organizational knowledge and supports
interoperability among existing operational systems.
The availability of semantical information is
fully-exploited by the Semantic-driven Restructuring
(SdR) module, located at the KM&D layer, which
supports selection and manipulation of basic EEs in
order to dynamically restructure them prior to the
application of process mining algorithms executed
by the Process Mining (PM) module of the KM&D
layer. As mentioned in Sect. 1, this strategy is meant
with the aim of straightforwardly applying process
mining techniques over EEs, while taking
advantages from the available background
knowledge. The latter restructuring approach is
illustrated in detail in Sect. 4. On the other hand,
novel pieces of knowledge, possibly captured in
models and patterns extracted by the PM module
and stored in the Discovered Process Models (DPM)
repository of the KM&D layer, can be further
integrated in the actual OKB by means of the
EFFECTIVE ANALYSIS OF FLEXIBLE COLLABORATION PROCESSES BY WAY OF ABSTRACTION AND
MINING TECHNIQUES
161