Our research in this project was organized in two complementary thrusts:
Several important conclusions can be drawn from the project:
- Studying the nature of the knowledge necessary for Concurrent Engineering, including issues of depth and breadth, with an emphasis on both the detailed design of packaging, and the early design phases where packaging and partitioning decisions are made, and
- Building on our previous work by addressing additional issues, including Abstraction, Negotiation, and Uncertainty.
- The Galileo language is capable of expressing both complex knowledge from a range of concurrent engineering disciplines relevant to design and the constraining influences exerted by these different aspects of the life cycle on each other.
- Perspectives, which are implemented directly using the field of view feature provided by Galileo, have been found to contribute greatly to simplifying user interaction with the advisor.
- Structuring data into complex objects has been found to greatly simplify the designer's task.
- Different users need to view their perspectives differently. For example, artifact designers usually prefer to use drawings while other disciplines find the spreadsheet or the relational database metaphor more appropriate. We have found that constraint networks can be treated as an underlying common representation for several user-level forms of representation.
- Constraint-based systems are easier to write, and are often an order of magnitude shorter than rule-based systems for the same task, but the problem of knowledge acquisition does not go away.