JESS Tutorial for free study group?

[Mike Van]

Many Jess Tutorials spend,
Too much time in the weeds,
A hand on tutorial is what I'd like to find,
To share with Study my group geeks.

The best lessons would begin,
With an overview in sizable slices,
Then would start taxing my brain,
With quizzes, coding, and practical exercises.

If anyone here has something like this,
I'd really appreciate it tons,
If you would post it or a link to it,
so my study-group can start having fun.

:-)

[Jason Menard]

Very lyrical, but I'm going to move this to Other Java Products and Servers where our Jess questions are usually answered. I think there's a guy there who knows a thing or two about it.

[Ernest Friedman-Hill]

Thanks for the poem! All of the resources I'm aware of are linked from the Jess web site. There's also the book "Jess in Action", of course; see the link below.

[Mike Van]

Originally posted by Ernest Friedman-Hill:
Thanks for the poem! All of the resources I'm aware of are linked from the Jess web site. There's also the book "Jess in Action", of course; see the link below.

Thanks Ernest, you're the man! My study group is looking specifically at how AI engines have been implemented in large companies who have the following issues:
1) a massive number of "facts" (using JESS terms); too many to be permanently instantiated as Java Objects (think in terms of millions at any given moment);
2) a millisecond turn-around time for complex rules.

The best example we've been able to find of this is the freight industry where there are millions of possible destinations a package could go to, but a "path" is quickly discerned before the package even gets on a truck.

From a very high-level point of view, if all "facts" are stored in a database, you'd have to do a SQL call to pull the unique facts for each package or check (name, destination, source, etc), create Fact objects in Java, run the rule, then destroy the Fact objects (make the available for destruction). Unfortunately, that algorithm seems quite wasteful.

Do you have any insight into this problem you can share?

Mike Van

[Mike Van]

Originally posted by Jason Menard:
Very lyrical, but I'm going to move this to Other Java Products and Servers where our Jess questions are usually answered. I think there's a guy there who knows a thing or two about it.

Really though? I've seen better in the meaningless drivel section...

[Ernest Friedman-Hill]

Originally posted by Mike Van:

From a very high-level point of view, if all "facts" are stored in a database, you'd have to do a SQL call to pull the unique facts for each package or check (name, destination, source, etc), create Fact objects in Java, run the rule, then destroy the Fact objects (make the available for destruction). Unfortunately, that algorithm seems quite wasteful.

That's more or less it, although Jess can be made to do the "pull from the database" part automatically, and doesn't necessarily require you to destroy the data immediately, and can do all of this multiple times in parallel.

Not sure what you mean by wasteful, really; any kind of Java-based processing at all, rule engine or not, is going to require pulling things out of the database into some kind of Java objects, working with them, then discarding the objects. The only alternative is doing all your business logic in stored procedures... *shudder*.