KADS-I models in CGs

Dr. Philippe A. MARTIN

This document is executable by WebKB-1. To try that, click here or use the first example in WebKB-1's Knowledge-based Information Retrieval/Handling Tool.

Abstract. This document shows how KADS1 models may be modelled using CGs, and how WebKB enables to mix and to associate CG representations and document elements. Hypertext navigation and conceptual queries may be done on the content of this document. This document could be completed to represent all the KADS1 interpretation models (the KADS1 generic task models that the knowledge engineer may adapt, complete and combine for modelling the problem solving expertise of its application). Then this document could be used as a way to store, to display, to structure and to document the KADS1 interpretation model library, and this library could be accessed by browsing or conceptual queries. The interpretation models of this document may be copied or refered in other documents for adapting them and combining them (this combinaison may eventually be guided by the maximal join function).

1 The models to build

In KADS, knowledge engineering is viewed as a process that bridges the gap between the required behaviour and a system that exhibits that behaviour, and this process is done through the creation of a set of models (Wielinga & al., 1992): the organisational model, the application model, the task model, the conceptual model and the design model.

$TC for Knowledge_engineering_with_KADS(x) are [Knowledge_engineering_with_KADS:*x]- { (Subtask)->[Environment_analysis]->(O)->[KADS1_Organisational_model] ->(SI-)->[Problem_analysis:*pa]; (Subtask)->[*pa]->(O)->[KADS1_Application_model]->(SI-)->[Task_analysis:*ta]; (Subtask)->[*ta]->(O)->[KADS1_Task_model]->(SI-)->[Function_analysis:*fa]; (Subtask)->[*fa]->(O)->[KADS1_Conceptual_model]->(SI-)->[Implementation_analysis:*ia]; (Subtask)->[*ia]->(O)->[KADS1_Design_model]; }$

The KADS-I Conceptual Model is composed of a "Model of Problem Solving Expertise", a "Model of Cooperation" and a "Model of Communication". In Martin (1993a, 1993b) we showed that the cooperation knowledge and the communication knowledge for an application may be complex and then the methods used for structuring and modelling the problem solving expertise should also be applied to the cooperation and communication expertises. Thus, we advice the construction of a "Model of Cooperation Expertise" and a "Model of Communication Expertise".

TC for KADS1_Conceptual_model(x) are

   [KADS1_Conceptual_model:*x]
      {
        (Part)->[KADS1_Model_of_Problem_Solving_Expertise];
        (Part)->[KADS1_Model_of_Communication_Expertise];
        (Part)->[KADS1_Model_of_Cooperation_Expertise];
      }

In KADS-I, a model of expertise is structured into 4 layers: the domain layer, the inference layer, the task layer and the strategy layer.

$NC for KADS1_Model_of_Expertise(x) are [KADS1_Model_of_Expertise:*x]- { (Part)->[KADS1_Strategy_layer]; (Part)->[KADS1_Inference_layer]; (Part)->[KADS1_Task_layer]; (Part)->[KADS1_Domain_layer]; }$

2 The generic task models

The generic task models include data-flow structures (the "Inference structures") and control structures (the "Task structures"). Inference structures also called interpretation models since they help to find which information is interesting in the expertise sources. They show the input-output relations between primitive tasks (called "inference") or complex tasks (called "task"), but do not specify the control on these tasks, i.e. in which order they must be applied (this is application-dependent).

The tasks modelled by the KADS-I generic task models are rather complex or precise and are hierarchically organised (the generic task models embody some methods to realize the tasks). The tasks modelled by the Common KADS generic task models are more general, and for each generic task model, various methods are proposed for realizing the tasks and then specialize the generic task model. That's why we consider that the Common KADS task are more general than KADS-1 tasks (see our hierarchy for KADS tasks). In our document on Common KADS, we represent in GCs the typical precedence relations that exist between Common KADS tasks.

2.1 The system analysis tasks

2.1.1 The classification tasks

The tasks of assessment, monitoring and diagnosis are classification tasks. There are several method for realising these tasks. Below is the inference structure for the systematic diagnosis. The systematic diagnosis consists in 1) the selection of a (sub-part) of the faulty system on the basis of a complaint, 2) a decomposition of some part of the system into a number of sub-components that play the role of hypothesis, 3) a prediction of a norm-value for a hypothesis (the norm is a value of a test which is consistent with the normal state of the hypothesis), 4) a selection of an observable, for which a value is to be obtained (the finding), 5) a comparison of the observed finding and the predicted norm.

$TC for KADS1_Systematic_diagnosis_task(x) are [KADS1_Systematic_diagnosis_task:*x]->(Descr)-> [Description_with_a_KADS_inference_structure: KADS1_L473_phmartin_AnyView [Difference]<-(O)<-[Compare]- { (DI-)<-[Finding]<-(O)<-[Select]- { (DI-)<-[Observable]; (SI-)<-[Hypothesis]<-(O)<-[Decompose]<-(SI-)<-[System_model:*m]; }; (DI-)<-[Norm]<-(O)<-[Specialize]<-(SI-)<-[*m]<-(O)<-[Select]<-(DI-)<-[Complaint]; } ]$

2.1.2 The prediction tasks

(Yet to be written).

2.2 The system modification tasks

The tasks of repair and remedy are system modification tasks.

(Yet to be written).

2.2 The system synthesis tasks

The tasks of configuration, scheduling, refinement design, transformational design, planning and open ended design are system synthesis tasks.

(Yet to be written).

2.2 The explanation tasks

In Martin (1993a, 1993b) we proposed an inference structure for the explanation tasks. Here is it in the CG formalism.

(Yet to be written).

3 References

Martin Ph. (1993a). Adaptation de KADS pour la construction de Systemes a Base de Connaissances explicatifs. Actes des 4emes Journees d'Acquisition des Connaissances (JAC-93), Saint-Raphael, March-April 1993.

Martin Ph. (1993b). A KADS refinement for Explanatory Knowledge Extraction and Modelling. Proceedings of the 6th Australian Joint Conference on Artificial Intelligence (AI'93), 16-19 November 1993.

Wielinga B., Schreiber G. & Breuker J. (1992). KADS: a modelling approach to knowledge engineering. In Knowledge Acquisition (1992) 4.