Beefy Boxes and Bandwidth Generously Provided by pair Networks
There's more than one way to do things

Re: Re: Re: Re: Rules-based Perl?

by Brovnik (Hermit)
on Jun 26, 2001 at 20:17 UTC ( #91643=note: print w/replies, xml ) Need Help??

in reply to Re: Re: Re: Rules-based Perl?
in thread Rules-based Perl?

I haven't seen any statement of purpose yet (or in XP jargon "Users stories").

Are you expecting the knowledge engineer (the person writing a rules base) to be a programmer ? A Perl programmer ?

What degree of complexitiy of (real-world) problem do you expect to be able to solve ?

For a user, a table isn't a complex concept, but inputing it might be unless the system is designed to cope at the beginning. Writing a rule such as (in my pseudo code) :

[ID in (A B C)] => [D = rows from TABLE_D where KEY=ID]

could be difficult otherwise.

Re. The rule efficiency, you are right, this is the hard part.
To some extent the responsibility lies with the knowledge engineer (read ruleset creator), since you can write bad (==slow) programs/algorithms in any language.

Giving the engineer tools (such as the priority / weighting mechanism you mention can help, as can other options such as choice of search mechanism and direction.

In my experience, a lot of knowledge engineer time is spent tuning the knowledge base, and the system should be proactive in helping that process, providing timing / coverage analysis.

Replies are listed 'Best First'.
Re: Re: Re: Re: Re: Rules-based Perl?
by Masem (Monsignor) on Jun 26, 2001 at 20:34 UTC
    At this point, it's merely for the purposes of "because it (the lack of perl&rules) is there". Mind you, yes, I'm thinking down the road and making this professional quality, which means that I have to consider the audience at that point. But the initial proof-of-concept would be a working rules engine, in which simply flow control is directed by what facts; anything beyond that , such as how facts are formatted, how rules are defined, etc, is mostly the 'user interface' part of the problem. And by providing a working engine, weaknesses and strengths of the interface can be figured out faster.

    I guess my overall goal is to provide a way to run a rules-like system (which in my previous experience has had weak interaction with the OS outside of input and output to terminal) using perl (which has strong OS interaction including networking, etc). An ultimate use would to place a perl script with the rules engine at the backend of a network socket, such that new facts can be introduced remotely and actions on local and remote systems taken as such (a so-called business logic server); so the language and depth would have to be simple enough to write the rules in but still allow for taking full advantage of using perl. There's obvious approaches to doing this, but determining how to draw the line between perl-isms and language elements I can introduce is way down the road.

    Once I get moving on this, I'll be posting the code as I've done with other modules of mine here for feedback and such, at least until it's stable and usable enough for CPAN inclusion.

    Dr. Michael K. Neylon - || "You've left the lens cap of your mind on again, Pinky" - The Brain

Log In?

What's my password?
Create A New User
Domain Nodelet?
Node Status?
node history
Node Type: note [id://91643]
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others cooling their heels in the Monastery: (3)
As of 2021-12-09 06:26 GMT
Find Nodes?
    Voting Booth?
    R or B?

    Results (36 votes). Check out past polls.