CoDE Publications CoDE Publications
IRIDIA Publications IRIDIA Publications
SMG Publications
WIT Publications
WIT Publications
SMG Publications
Home People Research Activities Publications Teaching Resources
By Class By Topic By Year
By Class By Topic By Year
login
D. Massart. La gestion de la complexité des schémas conceptuels à base d'objets. PhD thesis, Université libre de Bruxelles, Brussels, Belgium, April 2002.

Abstract

Information modeling consists of a set of techniques that aim at improving our understanding of the structure and functions of complex parts of the world, as they are perceived by our senses, by describing them in an abstract intensional manner. The parts of the world to be described are referred to as application domains, their descriptions as conceptual schemas, and the languages used to build schemas as conceptual models. Conceptual schemas are symbolic representations that are natural for the specialists of the underlying domain and that enable the formulation of various symbolic treatments considered as useful, e.g., to understand information, to transform it, to generalize the available information, to make assumptions and explore their consequences. One of the most significant barriers to the effective use of information modeling stems from the inability of current conceptual models to effectively help users manage the size and the complexity of large schemas. That is why various proposals were made to enrich conceptual models with complexity-management mechanisms. So far, their results have been rather disappointing. We define the complexity of a conceptual schema as the difficulty for human users to, e.g., construct, comprehend, and modify the schema diagram. The principal limits to those processes stem from the inability of the human brain to process large amounts of information. Managing the complexity of a schema thus consists in shielding its users from being confronted with too much information at the same time. We posit that modifying conceptual models to enrich them with mechanisms for complexity management hampers their usability. We think that building direct and faithful representations is the best recipe for schema simplicity and that managing their complexity is best left to a separate process. To support that thesis, we propose a method of complexity management that leaves the conceptual model unchanged and avoids exposing schema users to more information than they can comfortably manage as a single chunk. The method starts with a conceptual schema as it is produced by the activity of information modeling. The network of classes and their associations that makes up the schema is broken down into a backbone network and a collection of subnetworks. The backbone network provides an overall picture of the schema showing its main elements only (i.e., the classes corresponding to the most important concepts of the application domain and the shortest paths between them). The subnetworks give detailed views of the vicinity of each main class. Main classes are elicited using the number of associations in which they participate as a heuristic measure of their importance. The conceptual model is unchanged: no new abstraction mechanism is added, and no new notations or conventions are introduced. The schema is unchanged: the same classes remain connected by the same associations. It is supplemented with guidelines for reading its complexity. The process is semi-automatic to, at the same time, relieve users from complex schema-manipulation tasks and leave open the possibility for user validation and revision of the the results of each stage of the method before going to the next stage. Case studies validate the method. It easily applies to schemas built with various conceptual models (e.g., UML class diagrams, Entity-Relationship diagrams) and it easily integrates with object-oriented analysis methods. The method has also proven useful for managing other complex representations, such as large web maps and bibliometry charts.


Updated: 2017-03-27