[PowerLoom Forum] PowerLoom status

[Hans Chalupsky]

Cameron,

answers in-line:

>>>>> Cameron Ross <cross at symboticware.com> writes:

> 1) Are there thoughts on how many active projects employ PowerLoom in a
> significant way?

Unfortunately, I don't know.  My guess is that it is somewhere > 10
and < 30.  All I know are the downloads which are over a 1000 or so
plus occasional message flurries on the forum.

> 2) Are there currently any production-grade systems using PowerLoom?

Most probably not.  There are some projects that use it in
client/server environments, but I think they are still in the
demo/experimental stage.

> 3) Is PowerLoom ready for production use?

It depends on what exactly you need (or what you mean by "production
use").  I think if you can engineer the right environment around it,
and if you have good control over what knowledge gets loaded and what
queries will run, you can wind up with a system robust enough for
production use.  If it is a very open-ended environment, e.g., where
multiple users load their own KBs and ask their own queries, then
you'll likely run into problems.

> 4) If not, what would be involved in bringing the codebase to such a state?

Again, it depends on what you need.  Major to dos are proper
thread-safe multi-threading, completion of large-scale persistence,
better support of standards such as Common Logic and OWL, rule
compilation for better performance, various bugs and idiosyncrasies
with our inference engine, improved/updated user interface, and
testing testing testing.

> 5) Does a next generation follow-on project to PowerLoom make sense?  If so,
> is one planned?

There is nothing planned at the moment.  The major driver for
PowerLoom extensions are applications such as link discovery,
information integration, probabilistic reasoning.  It is always
tempting to start from scratch and do it all over again, cleaning it
all up and extending it into new directions.  However, since you
always need so much infrastructure for a full-function KR system,
you'd have to reimplement a lot of stuff and make so many design
decisions again before you get anywhere close to where PowerLoom is
today.  So my tendency so far has been to try to extend PowerLoom
incrementally wherever that's needed instead of doing a complete
rewrite.  The major other stumbling block is research funding which is
hard to impossible to get today for this kind of work.

> 6) What are the alternatives to using PowerLoom?

CLIPS, Jess, OpenCyc, Pellet, JTP, KM, Algernon, Allegro Graph,
Prolog, Jena, etc., again depending on what you need and what you can
tolerate.  I don't think there is a clear winner out there right now,
say comparable to something like Oracle or MySQL in the database world.

Personally, I'd be very interested to move PowerLoom into a more
commercial space, I just don't think there really is a market for it
(yet).

Hans

--------------------------------------------------------------------------
Hans Chalupsky, PhD                     USC Information Sciences Institute
Project Leader, Loom KR&R Group         4676 Admiralty Way
<hans at isi.edu>                          Marina del Rey, CA 90292
(310) 448-8745
--------------------------------------------------------------------------