Imagine your organisation's operations depended on the reasoning derived from a massive knowledge-base (say half a million semantic relationships between 100 000 concepts; 200 000 records in various "fact" tables; and half a million lines of some kind of knowledge scripting language like GELLO.)
Now imagine that knowledge-base was constantly evolving.
If your organisation is a tertiary hospital then I'm predicting your future will be something like the above within 10 years.
Which brings me back to meta-knowledege. What is it you are going to want track about the knowledge in your knowledge-base? This is my back of the envelope list:
Now imagine that knowledge-base was constantly evolving.
If your organisation is a tertiary hospital then I'm predicting your future will be something like the above within 10 years.
Which brings me back to meta-knowledege. What is it you are going to want track about the knowledge in your knowledge-base? This is my back of the envelope list:
- reliability – would you trust your life with this knowledge? logs of expert inspections, test-case reference, real-world feed-back and testing
- What authority or local expert(s) have inspected this knowledge have signed off on it? What test cases have validated the knowledge, and who has signed off on the test-cases?
- clarity – lack of ambiguity
- atomicity
- re-usability
- maintainability
- computational tractability
- context clarity – where does this knowledge apply? (is this universal, or does it just apply in this institution?) is any contextual limitation clear?
This comment has been removed by a blog administrator.
ReplyDelete