Published in: International Journal of Intelligent Systems, John Wiley & Sons. 1994, Vol. 9, n. 9, pp. 867-892

The VALID Project:
Goals, Development and Results[*]

Pedro Meseguer & Enric Plaza

IIIA - Institut d'Investigació en Intel.ligència Artificial,
CSIC - Spanish Scientific Research Council,
Campus Universitat Autonòma de Barcelona,
08193 Bellaterra, Catalonia, Spain.
Voice: +34 3 5809570;
Fax: +34 3 5809661;
{pedro | plaza} @iiia.csic.es

ABSTRACT: We recapitulate the work done in the Esprit-II VALID project, the aim of which was to develop methods an tools for the Validation of Knowledge-Based Systems. The project's goal was to undertake a comprehensive approach to the problem of Validation for existing KBS. In order to do so, several methods for different Validation issues were created, and different KBS were considered. The project concrete result is a Validation environment in which different KBS can be validated. It includes a Validation toolkit comprising seven tools, ranging from inconsistency detection to KBS inspectors, that the knowledge engineer can interact with. Validation tools produced by the VALID project have been tested on existing KBS and have been useful in improving their performance.

KEYWORDS: Verification, Validation, Testing, Expert Systems, Knowledge-Based Systems.

• 1. INTRODUCTION
• 2. PROJECT DEVELOPMENT
  • 2.1.WORKPACKAGE 1
    • 2.1.1. State of the Art
    • 2.1.2. General Formal Model
    • 2.1.3. Quality Methods, Tools and Plan Definition
    • 2.1.4. VETA Definition
  • 2.2. WORKPACKAGE 2
    • 2.2.1. Development of Validation Utilities
      • Cognitech's choices
      • CRI's choices
      • CEAB's choices
    • 2.2.2. VETA Implementation
    • 2.2.3. Interfacing with Existing KBS
    • 2.2.4. Quality Tools Implementation
  • 2.3. WORKPACKAGE 3
    • 2.3.1. KEE Feasibility Study
    • 2.3.2. Study of integration between different shells and VALID
    • 2.3.3. Evaluation of Validation Tools
    • 2.3.4. Final Packaging
• 3. ACHIEVED RESULTS
  • 3.1. Theoretical results
    • 3.1.1. Taxonomy
    • 3.1.2. Specification of the Validation Metalanguage
    • 3.1.3. Validation Techniques
  • 3.2. Methodological results
    • 3.2.1. Validating Expert Systems: Difficulties, Solutions and Tools
    • 3.2.2. A Quality Manual for KBS Development
    • 3.2.3. A Grid to classify and describe Validation Tools
  • 3.3. Software results
    • 3.3.1. VALID Architecture
    • 3.3.2. Common Conceptual Representation
    • 3.3.3. VETA
    • 3.3.4 Validation Tools and Techniques
      • CONKRET
      • CONTROLLER
      • CURRICULUM KB
      • EXECUTION-VISUALIZER
      • IN-DEPTH
      • PENIC
      • WITNESS
    • 3.3.5.VALID User Interface
  • 3.4. Experience results
    • 3.4.1. Learning on Validation
    • 3.4.2. Development of a General Formal Model
• 4. EVALUATION OF THE PROJECT AND PROJECT RESULTS
  • 4.1. EVALUATION OF THEORETICAL RESULTS
  • 4.2. EVALUATION OF THE METHODOLOGICAL RESULTS
  • 4.3. EVALUATION OF THE SOFTWARE RESULTS
    • 4.3.1. Evaluation of the Architecture
    • 4.3.2. Evaluation of CCR
    • 4.3.3. Evaluation of VETA
    • 4.3.4. Evaluation of validation Tools
    • 4.3.5. Evaluation of the user interface
• 5.CONCLUSIONS
• 6.REFERENCES