Newsgroups: rec.arts.int-fiction
Path: gmd.de!xlink.net!howland.reston.ans.net!pipex!uknet!pavo.csi.cam.ac.uk!rit10
From: rit10@cl.cam.ac.uk (Richard Tucker)
Subject: Re: Case based Int-Fiction parsers
Message-ID: <1993Nov17.141012.26959@infodev.cam.ac.uk>
Sender: news@infodev.cam.ac.uk (USENET news)
Nntp-Posting-Host: ouse.cl.cam.ac.uk
Organization: U of Cambridge Computer Lab, UK
References: <2c6igf$a31@mojo.eng.umd.edu> <2c9gi9$usa@msuinfo.cl.msu.edu> <1993Nov16.133234.15384@infodev.cam.ac.uk> <2cavig$ssp@msuinfo.cl.msu.edu>
Date: Wed, 17 Nov 1993 14:10:12 GMT
Lines: 91

In article <2cavig$ssp@msuinfo.cl.msu.edu>, dunham@cl-next4 (Steve Dunham) writes:
|> 
|> Richard Tucker (rit10@cl.cam.ac.uk) wrote:
|> : In article <2c9gi9$usa@msuinfo.cl.msu.edu>, dunham@cl-next4 (Steve Dunham) writes:
|> : |> 
|> : |> There are 3 ways to use verbs in English: Transitive, Intransitive,
|> : |> and Bitransitive.
|> 
|> : You mean 'ditransitive', I think.
|> 
|> It was called bitransitive in my Syntax class.
|> 

Perhaps this is a transatlantic difference; I got 'ditransitive' from a
syntax class too!

|> : |> 
|> : |> intransitive is when the verb doesn't take an object, e.g.
|> : |> 
|> : |> 	> leave
|> : |> 
|> : |> transitive is when the verb takes one object, the direct object, e.g.
|> : |> 
|> : |> 	> drop the box
|> : |> 
|> : |> bitransitive is when the verb takes two objects, the direct and
|> : |> indirect objects,
|> : |> 
|> : |> 	> give the dog a bone
|> : |> 
|> 
|> : These are indeed three verb types in English, but there are loads more
|> : not just using noun phrases after the verb, most of which are ignored
|> : by the average adventure parser. For example:
|> 
|> : a> paint the wall red           (NP + ADJ)
|> : b> turn the tv on               (NP + PART)
|> : c> curl up                      (PART)
|> : d> curl up small                (PART + ADJ)
|> : e> pretend to be dead           (INF)
|> : f> tell her to be quiet         (NP + INF)
|> : g> suggest going north to robot (ING + to + NP)
|> 
|> (c-d) yes, the participle, I believe it is the remnant of a construct
|> in Old-English analagous to that which became the seperable prefix in
|> German.  (It also appears in Old Norse.) I forgot to mention that, it
|> is quite important in an adventure parser.  

And (b) too, in which the verb and particle behave very like a German
separable verb. Semantically also, turn+on is really a different verb
from turn, but paint+red is the same verb as paint, just modified.

|> 
|> (f) I don't think this would be hard to parse (the infinitive clause)
|> and should definitely be parsed by adventure parsers; also, it should
|> let you use the vocative, as in Advsys, e.g.
|> 
|> 	Dog, take the key
|> 
|> (g) I don't know if this is really necessary, and it could be a pain
|> to parse... (what do you mean by (ING), present participle?) ...

Yes, sorry for my stupid abbreviations.

|> 
|> : etc, etc. I'd have thought a-d could be handles by modern IF, but e-d
|> : are perhaps too complicated.
|> 
|> Nah, it's not too complicated, its just challenging; that one of the
|> things that makes the design of adventure authoring systems fun. :)
|> 
|> Steve
|> dunham@gdl.msu.edu
|> 

Yes, the difficulty isn't in the parsing so much as in the world-modelling
of the adventure; I was thinking that if we allow (f), we're going to have
people typing things like

  > tell ford to tell vogon to shut up

whereas at the moment they don't try

  > ford, vogon, shut up

which would be the logical infocom-style equivalent. Still, I guess there's
no reason why the system shouldn't do speech acts between NPCs, it's just
that I can't think of any that do. Hopefully I'm wrong; it would certainly
make for a better adventure world.

Richard/
