Parallelism & Intelligence

Solution Evaluation

Current techniques
Current techniques for constructing information systems are mainly structured programming and OO, databases, $$ algorithms. They aim flexibility by clarity and above all reuse, by generality. But then a first important question arises: don?t we sacrifice performance for generality? Can we combine both or is it a tradeoff?
Secondly, do modern SE-techniques really guarantee the necessary reuse and flexibility?
Solution proposal: Optimal Code Generation !! (see LEVEL & PROJECT 7)

AI-techniques
When analysing the requirements of an ideal parallelisator, I will consider AI-techniques and evualuate them:
- expert system
- ontology
- NN
- GA
- Search (brute force)

Synthesis of my criticism:

Half solutions ($$ explain)
Instance solutions ($$ explain)

Requirements
- bereikbaarheid
- beschrijfbaarheid
- minimal information
- information system

Ontology
I have questions about:
- flexibility
- concept definition (is not)
- concept recognition
- different base hierarchies: a problem!?
- vision: we need answers to questions