Information Technology Reference
In-Depth Information
A context describes the requirements against which the schema is evaluated.
It is defined by a triple (A,P,D) ,where A is an abstraction level, P is a data
modeling paradigm and D is a set of quality requirements. The abstraction
level A and the paradigm P define the use of a specific model, e.g., UML class
diagrams at the logical level. According to level A , some constructs of P will
become undesired. For example, a conceptual ER-like schema should not include
constructs that explicitly or implicitly define a foreign key.
A rating is an ordering of the members of an equivalence class for a specific
context. It states the extent to which each member meets the quality criteria
D of the context. Different methods have been proposed to define these scores.
Collecting expert estimation is the preferred technique since it requires less effort
than standard empirical studies based on schema global evaluation. Ratings can
be used in several ways, notably (1) to compute metrics for the schema under
investigation and (2) to suggest schema improvement. A rating also allows to
identify in the equivalence class a best practice as the construct that has the
highest score for the context of the rating.
A quality method comprises analysis, evaluation and improvement methods.
An evaluation method provides global and detailed quality scores for the schema.
An improvement method is based on the replacement of constructs with a low
rating for a context by a better construct in its equivalence class, for instance
the best practice of the context.
4 Quality Requirements
Quality has become a major research field in software engineering, though its
scope, its objectives and its evaluation techniques have not gained sucient
consensus so far to consider it a mature domain. For example, similar but still
significantly different definitions of the very basic concept of understandability
can be found in [10], [11], [12] and [13]. Definitions have evolved with time
and standards have been proposed such as the ISO quality standard [12,14].
Unfortunately definitions available in a standard often appear too general and
not intuitive enough when addressing the quality of a specific software product.
This lack of precision also makes it uneasy to develop convincing operational
methodologies and to build supporting tools.
In this paper, we consider three essential qualities of conceptual schema con-
structs, namely simplicity, expressiveness and evolvability. In the following, we
provide definitions and interpretations of these non-functional requirements.
- Simplicity: Simplicity is a sub-characteristic of the understandability qual-
ity requirement in the ISO/IEC 9126 standard [12]. In [15], a schema is said
to be simple if it is constructed upon simple concepts . The notion of sim-
ple concept relies on a measure of the complexity, which is itself related to
the number of some specific elements, such as relationship types. Instead
we define the simplicity as the property of types of facts of the application
domain being represented as simply as possible . This definition encompasses
the notion of minimality and low complexity. A practical definition could
Search WWH ::




Custom Search