Title: Ontology based framework for System Engineering Knowledge Capitalization and Reuse

Abstract

•\tAuthors: -\tPrimary author: Olfa CHOURABI *, PHD student, part time teacher at CNAMuniversity. CEDRIC laboratory/RIADI laboratory**. Email:olfa.chourabi(at)voila.fr-\tYann POLLET*, Professor, System integration department, CNAM, Paris *: chaire d’intégration de systèmes, 291, rue saint martin, PARIS** : campus universitaire de la Manouba, TUNISIA.•\tPrimary Track : 1. Knowledge Management; Secondary Track: 2. R&DManagement•\tThe abstract is (i) abstract for a paper•\tWould like this abstract/ paper to be considered for (i) Studentcompetition paperKey-words: System Engineering, Knowledge Management System, Ontology,Knowledge Representation, Conceptual GraphsAbstract:System Engineering (SE) is an interdisciplinary approach to enable therealization of successful systems. It is defined as an iterative problemsolving process aiming at transforming user’s requirements into asolution satisfying the constraints of: functionality, cost, time andquality.SE processes comprise highly creative and knowledge-intensive tasks thatinvolve extensive problem solving and decision making activities amonginterdisciplinary teams. System engineering projects involve the definitionof multiple artifacts, that present different formalization degrees, suchas requirements specification, system architecture, hardware and softwarecomponents. Transitions between the project phases stem from decisionmaking processes supported both by generally available domain and designknowledge. All these elements along with “best practices” and engineers“know-how” form implicit knowledge assets not well valorized nowadaysin SE organizations. In such context, traditional information managementsystems fail to provide adequate support due to the lack of formalizationof their fine-grained process knowledge. Process support should thereforealleviate this shortcoming by eliciting new engineering ways of working andimproving the existing ones based on substantial experiences capturedduring process execution.We argue that Knowledge about engineering processes constitutes one of themost valuable assets of a SE organization. Normally, this knowledge is onlyknown implicitly, relying heavily on the personal experience background ofeach system engineer. To fully exploit this intellectual capital, it mustbe made explicit and shared among system engineers. Consistent andcomprehensive knowledge management methods need to be applied to captureand integrate the individual knowledge items emerging in the course of asystem engineering project.This paper is a part of an ongoing research project that aims to developan environment for knowledge capitalization and reuse in SE community. Ourresearch is close to the concept of project memory. The key idea behind ourapproach is a flexible ontology-based schema with formally definedsemantics to enables the capture and reuse of engineering experience.Our environment is based on a set of consensual ontology’s of theengineering domain where entities such as actors, processes, artefacts, areinterlinked to represent the knowledge essence of the domain.More precisely we propose an ontological framework organized into foursemantic layers: the meta-meta model layer, aims to describe super-conceptsthat are the same across all domains, it corresponds to top-level ontologyor foundational ontology. The meta-model layer defines concepts andsemantic relations characterizing system engineering processes, itcorresponds to “SE core ontology” (specific to system engineering anddomain independent). In the model layer, application domain dependantontology’s, such as software engineering, Aeronautics etc., are createdusing Meta model primitives. The fourth layer corresponds to semanticannotation on SE project assets defined using conceptual vocabulary fromthe model layer. In this way, all SE project assets are captured as formalknowledge models, by instantiating these ontological concepts.A key issue in defining this knowledge management support is the adequatechoice of knowledge representation formalism as well as its underlyingreasoning capabilities. We choose conceptual graphs as a knowledgerepresentation formalism for capturing SE project assets. We introduce anextension to conceptual graphs which allows the representation of theinherent variability in a system engineering situation. We defineStructural Graphs (s-graphs) as a Conceptual Graph with a combinatorialstructure on its concepts and relation nodes. We show how s-graphs cancapture sets of possible knowledge models in a system engineeringsituation, as opposed to static, “snapshot” facts about the world. Wemotivate the need for such representation power and provide examples.Logical semantics of the s-graphs and the associated operations are alsodiscussedThe paper is structured as following: the first section presents ourmotivations for supporting SE Community with a Knowledge Management System.Based on existing approaches and industrial needs, we propose a KMFramework for SE community in section 2. In Section 3, an illustrativeexample is provided. Finally significant related works are discussed andsome future research directions are outlined;