[MUD-Dev] Text Parsing

[Ola Fosheim Grøstad <olag@ifi.uio.no>]

Nathan F Yospe wrote:
> Am I the only one here who uses a NLP for *output*? This is a legitimate
> use of a parser, and as far as I know, I'm the only one attempting quite
> so ambitious an undertaking. 

Yes, you are probably the only one with sufficient hubris. *grin* But
grammars used for generating output has been discussed before of course.

> Others have mentioned markups and inserting
> weather or time affected variances, as well as attention and focus based
> organization of clips, but ... who else has actually started this? There
> are several things I'd like to hash out with someone...

I haven't started anything. I have not even decided upon the general
representation of the world. However, everything that is stored will have a
formal representation beyond text, so I will have to generate output
somehow. This I have thought some about, and I probably will start with the
client, delaying the representation of the world another six months or more.

One basic problem is to support speech devices and their low throughput as
well as not telling users with graphical clients what they already can see.
This means that there will have to be some clever filtering. That means the
system will have to know what the user is very likely to know already. That
means I have to stuff some of this on the client side for storage reasons
and in order to get better response times.

Basically I think about conceptually building a graph representing what the
system could present the user, and then prune it based upon dependencies and
priorities. For actual output, I don't know. I assume that it will either be
some kind of template based approach (premade patterns that are matched) and
maybe some rewriting system. Descriptions are likely to become inane, so I
won't start on this in a long long time, and will look at other fields later
this summer.

Expected problems: descriptions will become long and boring, or they will
become incomprehensible or not contain what the user actually are interested
in.

My agenda for July: finding or building survey of story generation research,
reading up on OO-Prolog and similar tools, survey of autonomous agents and
emotions and relevant HCI research.

Is there anything else I ought to look at?
--
Ola Fosheim Groestad,Norway      http://www.stud.ifi.uio.no/~olag/



_______________________________________________
MUD-Dev maillist  -  MUD-Dev at kanga.nu
http://www.kanga.nu/lists/listinfo/mud-dev