307 
resources. Too late introduction of new techniques may cause 
substantial economic losses, whereas too early introduction can lead 
to failures. Thus assessment of the operational maturity of the new 
techniques is important. 
On the other hand, the knowledge gained from experience in 
production provides an invaluable feedback for further development 
and optimization of the techniques. 
2.4 Techniques for GI applications 
These techniques represent a significant part of the state of the 
art in the GI users' domains. The specific GI should be balanced 
against the users' problem-solving models, i.e., it should be 
conditioned for the processes in the specific user's domain. This 
can be achieved by mutually adapting the techniques for GI 
production and those for GI use. The production and use of GI are 
linked by the corresponding specifications. Hence, the specifying 
process should be strongly interactive between the users and 
producers of GI, and iterative from coarse to fine. 
3. CRITERIA 
The criteria for evaluation emerge from the objectives and context 
of the two techniques or system types. Most essential is to identify 
the significant criteria and to quantify their importance. They can 
be structured into the overall criteria and the institution-specific 
ones; the latter change from one institution to another, and are 
beyond the scope here. 
The overall criteria address general technical features and economic 
factors, such as: 
- Versatility: diverse input, procedures and GI products; 
- Flexibility: ability to change, expand, integrate, modularity, 
compatibility; 
- Cost of system acquisition, operation, support; 
- Performance of components, procedures, and GI products; 
- Reliability of components, procedures, and GI products; 
- Time efficiency of processes and operations; 
- Automation: degree, suitability; 
- Interactive capability and human factors; sensory, processing, 
control; 
- Support for training, maintenance, upgrading, etc. 
- System introduction in the production environment, etc. 
Each of these main classes can be subdivided into more specific 
criteria. They can be structured hierarchically and weighted. 
Accordingly, a value model can be formulated, which is beyond the 
scope here (Me Kenzie, 1980). 
Evaluation may pertain to a technique or a system as a whole, or to 
its individual parts, such as the hardware components, software 
modules, process stages, GI products, and support. The criteria and 
thus the value model should reflect the aim of the evaluation.