SEKE 2012 Proceedings - Knowledge Systems Institute

identification, classification, problem, context,
solution, role, artifact, relationship, guidance,
management and evaluation [12]. In addition, these are
differently covered by process patterns descriptions as
it is shown in Fig. 2 most of them pay attention to the
four main facets: context, solution, problem and
relationships (15%, 14% and 13%).
The lack of terminological consent refers to the
problems of polysemy and synonymy addressed in
labels used to describe patterns. Indeed, we find terms
such as Consequences used to express a Resulting
Context in PPL as well as a Guideline in Gnatz.
Moreover, others different terms are being used to
address the same concept such as Intention in
RHODES to describe a pattern Problem, instead, the
term Intent is used in Störrle.
B. Unified Conceptualization of Process Patterns
To overcome these lacks, a first step was to create a unified
conceptualization of process patterns. Mappings efforts [12]
were necessary to achieve this goal leading to a process
patterns’ meta-model unifying patterns knowledge
representations.
Fig. 3 provides an abstract view of the meta-model
structure in which we consider a process pattern information
description from six facets:
The identification facet encapsulates a set of properties
identifying a pa ttern such as pattern name, author(s),
keywords, pattern’s classification (type, category,
abstraction level, and aspect) as well as pattern origin
(project and participants) and pattern artifacts (used
and/or produced).
The core information is the main pattern facet
embodying details about the well-known triplet:
problem, context and solution.
The relationships facet expresses how a pattern could
interact with other patterns (e.g. similar patterns,
refinement patterns, subsequent patterns, and antipatterns)
The guidance facet refers to the support level provided
by a pattern to be comprehended and used (e.g. known
uses, example, literature, illustration, etc.)
The evaluation facet provides feedbacks on pattern
application (e.g. discussion, confidence, maturity, etc.)
The management facet provides general information
about a given pattern (e.g. version, creation-date)
In order to validate the proposed meta-model, we adopted
an ontology-based approach providing common and shared
architecture and terminology for better patterns’ capitalization
and reuse. The proposed ontology [12], named MetaProPOS
and acronym for Meta Process Patterns’ Ontology for Software
Development, aims to unify all th e proposed process patterns
descriptions, providing thus semantic interconnection and
composition of relevant and unified bodies of patterns
knowledge.
Artifact
4%
Role
6%
Guidance
11%
Relationship
13%
Identification
11%
Figure 2. Architectural dissent in patterns descriptions [12].
Figure 3.
Evaluation
3%
Management
2%
Abstract view of a unified process pattern.
Because of space limitations, we could not give details of
all the modeled facets. So we choose to illustrate this work by
the relationships facet which is quiet important for the pattern
reuse process. Fig. 4 shows the relationship hierarchy
considered by the proposed meta-model. An alternative pattern
is a pattern that could be applied as an alternative solution
which can be different from a similar pattern. A refinement
pattern should give details to an a bstract one. A consequent
pattern is a pattern that is implied or that could be applied after
a given pattern. An anti-pattern is a pattern that could not be
applied with another given pattern.
Figure 4. The relationship’s facet.
Problem
13%
Classification
8%
Context
15%
Solution
14%
507