Configuration Issues
This list of issues was presented at the end of the AAAI 1996 Fall Symposium Configuration Workshop to stimulate discussion and to serve as a potential list of topics for a future workshop.
- What distinguishes Design from Config?
- What are the "extreme cases of config?
- e.g., Sorting?
- Where do config concepts break down?
- e.g., where and what are the ports in a pile of blocks?
- Are there subclasses of config?
- With different sized search spaces?
- e.g., use all given components vs. some
- e.g., objects with parameters vs. not
- e.g., continuous values for parameters vs. discrete
- Are the methods we use equivalent in power?
- If so, why does it matter what we use?
- What kinds of knowledge about a config. domain can we take advantage of?
- ... to improve Solutions (e.g., for optimality)
- ... to improve process (i.e., for efficiency)
- (eg. abstractions & taxonomies;
- templates -- association & arrangement)
- How much, and when, do user concerns/demands/needs affect the configuration process, as opposed to just affecting the interface.
- Config for X
- (X = assembly, bundling, ...)
- Do we need new techniques?
- Why is Description Logic suitable for configuration problems?
- Some comments (Dave Brown).
- A response (Tony Weida).
- Why is Constraint Satisfaction suitable for configuration problems?
Other issues from Rainer Weigel:
- What are the ingredients for a "flexible" configuration system?
- What are the requirements for a "maintainable knowledge representation" for configuration systems?
- Where do the "real" difference between DL and CSPs for configuration come from?
- What are the research topics from the industrial point of view?