Call for Papers (.pdf) (.txt)

Organizers

Petko Valtchev
DIRO, Université de Montréal
Canada
e-mail: petko.valtchev@umontreal.ca
phone: +1 (514) 343 7599
(primary contact)

Hernan Astudillo
Financial Systems Architects, New York
USA
e-mail:hernan@acm.org

Marianne Huchard
LIRMM-Université Montpellier 2
France
e-mail: huchard@lirmm.fr

Program Committee

Montreal Tourist Information

• Airport Survival Guide
• Downtown Hotels
• Excursions and City Tours
• Montreal Museums
• Montreal Exhibitions and Trade Shows
• Virtual Tours
  

Motivation

In object-oriented (OO) approaches (e.g., modeling, programming, databases, knowledge acquisition/discovery and knowledge representation), the core of a system are specialization hierarchies that organize the abstractions manipulated by the OO system. These abstractions, represented with classes, interfaces and types, can be used to represent concepts of the application domain and/or software artifacts required to support computation.

In OO knowledge representation and discovery, prevails the modeling purpose of classes; thus, hierarchies are repositories of validated knowledge, and support reasoning mechanisms (inductive or deductive) for automatic acquisition of new knowledge. In OO programming, the main purpose of hierarchies is organizing software components (classes, methods, etc.) to enable and ease their retrieval, use and reuse; thus, class hierarchies are implemented with inheritance, which also supports sharing and reuse of features (specification and/or code). In OO databases, both aspects (domain representation and component organization) co-exist, reflecting the modelling and the implementation view respectively.

Since OO analysis and design encompasses key aspects of domain knowledge acquisition and representation while strongly based on both programming and database modelling, the interpretation of OOA&D hierarchies shifts during design from a modelling use to an implementation one. Hence, many modern OOA&D methods support the gradual evolution of class hierarchies from fitness for one use to fitness for the other, usually by enabling creation of initial hierarchies within the application domain model, and then allowing them to "evolve" (or derive) into hierarchies for the computational environment.

Despite their wide and longstanding use in these disciplines, there is no standard methodology to construct and maintain a specialization hierarchy independently from the domain it represents and the modelling/programming artifacts it organizes. Unsurprisingly, hierarchies still give rise to questionnable interpretations and implementations. The manipulations of OO hierarchies are further complicated by their size, the variety and inconsistency of classification and generalization criteria used in their construction, and the evolution of the hierarchy that may reflect a shift in domain understanding and/or the natural dynamics of the domain itself.

Objective and Scope

The focus of MASPEGHI'03 is on the evolution of OO hierarchies in modern systems from a domain representation to an artifact manipulation view, including topics such as support for advanced hierachy manipulations (such as dynamic changes in class membership), integration of specialization/generalization hierarchies coming from different sources, and slicing of an existing global hierarchy to provide partial views to fit specific purposes.

The aim of the workshop is to bring together people interested in specialization hierarchy design, implementation and use, to summarize the state of the art in the field (current practices and tools) and discuss open questions.

Different approaches to the hierarchy management field(s) may be considered, such as:

• proposition of overall quality criteria and/or methodological guidelines for specialization hierarchy elaboration,
• definition and/or experimental study of hierarchy evaluation methods (e.g. metrics),
• design of automatic or semi-automatic tools for specialization hierarchy management (e.g., construction/evolution/re-engineering).

• limitations of modeling language expressiveness/semantics (e.g., pure object/property model vs. object/property/association model),
• ontological and implementational aspects of a specialisation hierarchy,
  
• quantitative approaches (metrics) to hierarchy quality evaluation,
• methodologies for hierarchy comparison,
• partially ordered structures (trees, DAGs, lattices, etc.) for representing speciallization hierarchies,
• transformation of specialization hierarchies during life-cycle phases (analysis, design, implementation),
• integration of hierarchies from different sources,
• multiple hierarchy management, the separation of concerns,
• discovery of class hierarchies from unstructured artifacts (e.g., flat set of classes, legacy procedural code, etc.),
• relevant results from connected domains such as knowledge representation and acquisition/discovery, statistical classification, etc.
• automatic tools for hierarchy management,
• lessons learned from automatic/manual specialization hierarchy construction,
• case studies.

Workshop History

• This is the second edition of MASPEGHI. The first edition was held in Montpellier, France, and was affiliated with  OOIS 2002 .
  
• A similar workshop, The Inheritance Workshop was independently held at  ECOOP 2002.
  

Workshop Location

Workshop Program (preliminary)

Workshop Registration

The workshop fees of 100 CAN$ include : a copy of the workshop proceedings, admission to the technical sessions, refreshment breaks, lunch and workshop reception on October
6th.
The registration is available via the ASE'03 registration site here.