home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!noc.near.net!hri.com!spool.mu.edu!uwm.edu!rpi!utcsri!relay.cs.toronto.edu!smoke.cs.toronto.edu!cs.toronto.edu!vinay
- From: vinay@cs.toronto.edu ("Vinay K. Chaudhri")
- Newsgroups: comp.ai
- Subject: Re: Questions on Concurrency Control for Knowledge Bases
- Message-ID: <92Dec20.140658edt.106@smoke.cs.toronto.edu>
- Date: 20 Dec 92 19:07:35 GMT
- References: <1992Dec20.183502.17178@dragon.acadiau.ca>
- Organization: Department of Computer Science, University of Toronto
- Lines: 44
-
- In article <1992Dec20.183502.17178@dragon.acadiau.ca> 881244y@dragon.acadiau.ca (Choon Lin Yeo) writes:
- >
- >Hi ,
- >
- > I have a few questions on the topic of CC methods for
- >Knowledge Bases:
- >
- >1) What are the methods used for CC in the existing commercial
- > knowledge bases? Example of how they are applied in this context.
- >
-
- Commercial knowledge bases (aka expert system shells) *do not* support
- multiple users. If they do, they process the requests of one user at a
- time.
-
- >2) Are Timestamp Ordering (TO) and Two Phase Locking protocols feasible
- > methods?
-
- As an initial solution, either of them can be implemented in a knowledge
- base. In the long term, I think, we need something better. May be,
- something, that uses the explicit semantic knowledge in a knowledge base.
- >
- >3) Which of these methods give a better performance in the context of
- > knowledge base?
- >
-
- This question has not been answered even in the context of databases, so
- I do not expect any easy answers in the case of knowledge bases. When
- there is not much data contention in an application, it does not really
- matter which algorithm you use. When there is heavy contention but you
- have unlimited resources (hardware and disks), then time stamp method
- (or optimistic methods are better). In practice, you never have unlimited
- resources, and under those conditions, locking performs better.
-
- The conventional wisdom is that *Locking Works*. Most successful commercial
- database systems use locking. Locking is a commercially proven technology.
- Therefore, I recommend locking based methods as a starting point. This is
- not to say that other methods such as version-based methods are not interesting,
- but the point is that for the other methods, more work needs to be done
- before they can be used in a knowledge base.
-
-
- vinay.
-
-