[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: a propos of artificial intelligence, linguistics, and geckos








That nice Mr Lenzo wrote:
> We need to discuss the whole thing; right now it's a pretty
> simple forwarding. Addi's repro on alternate networks, and Simon's
> mods(!) are part of the Field.

Right. As most of you probably know, I'm Very Busy at the
moment, and, although I can find time to work on 0.4x, I don't
really have the time or the energy to grok the new architecture,
much though I'd like to.

I'm happy, if it would be useful, to take over collecting patches
on the old stuff for the time being, so that development on the new
stuff will go ahead; my current schedule means that I may well be
up to helping out with Geckobot in August. At some point, I might
start testing out the Highly Dangerous Development patches, but I
can't promise.

On the other hand, I'm more than happy to bounce ideas around
and I'd be especially interested in the linguistic side of it
as well, both the parsing and the internationalisation.

That said, there are a few things that think would be a good
idea at the design level, but they're so obvious you've probably
thought of them anyway:

     The ability to add a filter to any stage of input/output.

     Complete abstraction of input/output for flexibility.

     Communication between components: when trying to answer
a question, we may need to head off to a web search component
to find information about the topic, which may in turn fire
off a dictionary component to help with understanding something
on the returned web page, and so on.

     Some state keeping would be nice, too. I recently had
to do some very, very crufty code for the infobot to get
information out of another IRC bot - it would save the original
questioner, channel and message type away (is this going to
be a Query object? Can they go on a stack?) and message the
other bot, then return to ordinary processing, special-casing
replies from the bot to relay onto the original questioner.
Yuck. If we can think of a way around that, it would be helpful.

We're also seeing the communication aspect coming up, which is
an important issue. (Are we starting to model social patterns?)
What do *you* do when you need to find something out? You
either already know it (it's in your DBM, as it were) or you
can ask someone else, you can look it up on the web or in an
encyclopaedia. In some situations, you may think you've got the
answer, but it doesn't seem to fit the question somehow. A
human can tell pretty easily when he's got the wrong reference
source for a topic; the query has a series of mental
associations. When we're looking up quavers, we can usually
work out if the answer's supposed to relate to music or cheesy
potato chips. We're about to hit a really nasty knowledge
representation issue here.

(Should we be talking to the perl-ai list about this too?)

> i've been told at CMU i can work on "anything i want to do",
> so i will be at least partly on refocussing on the 'bot --

Sounds a pretty good brief.

> I have speech in and out working,

Genius. That's a huge psychological boost for human-computer
interaction. I wonder where the project currently sits, in
terms of the chatting-with-machines world.

>From a Perl evangelism point of view, I'd really appreciate
it if someone could find the time to XS up festival and
sphinx. The more Perl this thing is, the bigger a propaganda
coup. :)

While we're being a propos of nothing, I'd just like to comment
on how ironic life is. About a year ago, my rival brought an
infobot onto the IRC network of my local hacker community.
Of course, I already had a good emotional reason for disliking
the bot, and it didn't help that it only knew about twenty or
thirty factoids, and answered everything with `bugger all, I
dunno.' What a useless piece of software, I thought. Now look
what I'm doing. Maybe it's spite. :) (Of course, the real
beauty is that the software hasn't changed, just the size of
the knowledge base I'm used to working with.)

> YAPC is a part of that, and other directions; let's look
> at international venues, also.

This was recently disgust(sic) on clpm, so I funneled it off
to a mailing list:
echo | mail european-yapc-subscribe@othersideofthe.earth.li
to join.

> i'm on the verge of proposing a long talk at TPC on
> particle, wave, and field in the world of the bot -- but
> as a situated talk on perl and "we."

By all means reach for the stars, but never forget to keep
your feet on the ground. When it comes down to it, only
the code matters.

Simon