Accepted:
00/00/1994
| |
/vol01/finega01/ | © Copyright 1994 | |||
| Volume 1 | Received: Accepted: |
00/00/1994 00/00/1994 |
|||
Andrew Finegan
RMIT Centre for Remote Sensing
Department of Land Information
Royal Melbourne Institute of Technology
GPO Box 2476V, Melbourne Vic. 3001, Australia
Email: rfeadf@minyos.xx.rmit.oz.au
The traditional systems approach to problem solving is based on the technique of reductionism, which solves a problem by fragmentation, one stage at a time. This technique is appropriate in complex and highly structured situations that are able to be well defined, particularly in terms of inputs and outputs. In information systems, this is formalized in the system development life cycle (SDLC) and the alternative method of prototyping.
Studies in knowledge elicitation by Gaines and Shaw [2][1], Shaw and Gaines [3], and Shaw [4] have focussed upon the need to use systemic and psychological foundations to develop models of human knowledge representation, acquisition and processing. It is argued that the standard formal logic of the accepted reductionist or mathematical systems theory may be inappropriate for knowledge elicitation, and Soft Systems Methodology (SSM) is identified as providing a suitable theoretical framework.
The builders of expert systems have generally adopted prototyping as the preferred method for system development. However, there is strong evidence that prototyping is of limited success, particularly where human factors and poorly defined complexity exist. There is the danger that prototyping can be "technology-driven", which can lead to the situation described by Stowell and West [p. 674]sto89:
... techniques which are "technology driven", such as prototyping, are likely to impress their own structure upon the knowledge elicitation process through the requirements of the technique's design and usage. It is important, however, that the expert is prevented from "trimming" his knowledge to fit the KE (Knowledge Engineering) method and the knowledge structure of the expert system, a difficult task to fulfil in the light of current KE practices.
They propose that a more heuristic and subjective approach should be taken for knowledge elicitation in complex and poorly defined problem areas, and suggest that "a possible candidate for this task is the system-based problem-solving methodology, SSM" [p. 676]sto89.
Similarly, the pitfalls of expert system development as described by Curtis [pp. 536-537]cur89 include such human factors as "the system does not match the working practices of the final users", "there is no willing expert(s) prepared to part with the knowledge essential for the expert system", and "management commitment is lacking or unrealistic". This argument is developed by Rodger and Edwards [p. 1071]rod89, who in examining the development of expert systems find that "from the problem-driven stance, the concept of `prototype' is problematic ...," and that prototyping "... is not suited to the solution of messier, real-world problems ....". This study also concludes that the alternative approach of Soft Systems Methodology is appropriate for the development of problem-driven expert systems.
A systemic approach to problem-solving is provided in a methodology developed by Peter Checkland, Professor of Systems at Lancaster University [8]. This is known as the Soft Systems Methodology (SSM). The concepts were developed through practical application and experience in a wide variety of complex managerial systems. The methodology is designed to allow the human element of such systems to be incorporated into system design work. It is not easily assimilated or applied, and its apparent simplicity may be deceptive. It may be used to analyze any problem or situation, but it is most appropriate for the analysis of systems that are not well defined.
The Soft Systems Methodology is described by Wilson [p. 64]wil84 as "a seven stage process of analysis which uses the concept of a human activity as a means of getting from finding out about the situation to taking action to improve the situation". These seven stages are illustrated in Figure 1.
Figure 1: The Soft Systems Methodology. Adapted from Checkland [8, p.163]
This figure represents the pattern of activities in the methodology, it does not necessarily impose a sequence in which it should be applied. As Wilson says: "The analyst may start with any activity, progress in any direction, and use significant iteration at any stage" [p. 64]wil84. The line between the real world and the systems thinking defines the boundary between the use of everyday language and systems language.
The Soft Systems approach is an evolving methodology that has been steadily developed into a systemic process of enquiry structured around a comparison between a real-world problem situation and conceptual models of relevant systems of purposeful activity [10].
Figure 2 presents a model for the application of Soft Systems Methodology as an iterative cycle of action research.
The case study focuses upon the issue of effective technology transfer. It is a systems investigation that is examining the technology transfer of remote sensing in Australia, with the objective of providing a better understanding of the management of a complex technology within human activity systems. It clearly illustrates that not only are there technological issues, but there are also challenges associated with organizational structure and efficiency, training and education, and government coordination.
An Australian report [11] indicates that successful commercialization of new technologies is dependent upon the linkages that exist between the public sector research organizations and industry. This emphasis on linkages is supported by the findings of Prager and Omenn [12] and Boyle [13] and is expressed by the following questions:
The negative perceptions of the last question include:
Studies of the technology transfer process for remote sensing have been undertaken by Ferns and Hieronimus [14], Forster [15], and Specter [16]. A report of particular relevance to this study is that of the Australian Space Office [17] which identifies the weaknesses in the commercialization of remote sensing in Australia. Analysis of these reports has identified common problem areas in remote sensing technology transfer [18].
The problems associated with technology management and technology transfer are complex, unstructured and poorly defined. This premise is supported by Flood's [19] use of technology transfer to illustrate the concept of situational complexity. Soft Systems Methodology has been used to provide the theoretical framework for the study of the processes of remote sensing technology transfer in Australia [20][18]. The design of an expert system for remote sensing technology management, based on the model developed by this study, is being developed as a practical outcome.
The first two stages of Soft Systems Methodology involve the examination of the background of the problem. This is expressed the form of the "Rich Picture" (Figure 3) which aims to show the elements of slow-to-change structure and elements of constantly-changing process within the situation being investigated.
The Rich Picture can be applied to the initial stages of the knowledge elicitation process to help develop a representation of relevant domains, and an understanding of the views of people within each domain. Stowell and West [21] suggest that the Rich Picture is very useful as a summary of the knowledge elicited from the expert. The analyst can use it as a prompt for discussions with experts, as an aid for assimilating knowledge elicited, and as a means of identifying the areas in which knowledge is limited.
In this stage a choice is made of relevant systems that the analyst believes will produce insight into the problem situation. The chosen systems are expressed in statements as the Root Definitions, which incorporate the points of view that make the activities and performance of the systems meaningful. The initial Root Definition for this study of technology transfer has been formulated as follows:
An industry driven system operating within research centres with the objective of transferring untransferred technology by: knowing about untransferred technology, knowing about targeted industries, selecting technology to be transferred, selecting means of transferring technology, applying those means to an industry, stimulating the ongoing transfer, and monitoring the success of such transfers; in order to benefit all involved parties, in an environment of research, industrial competitiveness, and national and international economic development.
The formulation of "good" Root Definitions is decisive to the creation of the conceptual model in Stage 4. Therefore, the Root Definition is tested against a group of elements known by the mnemonic CATWOE, that defines a check-list for Customer, Actors, Transformation process, Weltanschauung (worldview), Owner, and Environment. Invoking the CATWOE for this study results in:
The elements of CATWOE emphasize the need for what Shaw [4] terms constructive alternativism: that it is important to examine the problem from a number of viewpoints. The Root Definition and CATWOE provide the analyst with a framework for ensuring that all points of view and interest are considered in the knowledge elicitation process.
This stage is where a logical expansion of the Root Definition is made into the minimum necessary set of activities to define what the system actually does at a particular resolution level. The qualitative modeling process uses pictures and diagrams to define and communicate structure, logic, ideas and relationships. The Conceptual Model should be expressed by verbs.
The logical expansion of the Root Definition for technology transfer results in a Conceptual Model of three subsystems - `"nowledge", "criteria" and "application" - while the activity "monitor and control" remains at the first level of resolution (Figure 4).
Figure 4: Conceptual model.
This detailed model represents a human activity system that
can now be used to create a well-structured evaluation of the
state of the real world. This is achieved by comparing the
model with perceptions of "what is the present mechanism".
It provides a means of enquiring into areas of expertise which
seem difficult to understand or that have been poorly defined
by the expert [21].
Comparison of the Conceptual Model with the real world is
undertaken by comparing each of the second resolution
activities within the model with the real world problem
situation. This was achieved in this study by the rigorous
interviewing of project managers in agencies and companies
that use remotely sensed data. In the interview the following
questions were asked for each activity:
The study is now at this stage of the analysis.
The participative nature and strong focus upon human activity
systems of this methodology has facilitated the development
and testing of a systems model of a "messy", poorly defined
and complex problem area.
The use of the model as a knowledge elicitation tool has been
successful, both in the quality of the information gathered,
and in the response of the participants interviewed. It is
pertinent to note that the majority of individuals claimed
that their participation in the analysis has led to useful
insights into problems they are having with remote sensing
technology management. Many have volunteered to take part in
further studies.
The case study illustrates the application of Soft Systems
Methodology to the problem of remote sensing technology
management, and suggests that this approach is a suitable
method for knowledge elicitation in expert system development.
Stage 5: comparison
Conclusion
References
