home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!snorkelwacker.mit.edu!ai-lab!xbar!dmb
- From: dmb@xbar.ai.mit.edu (David Baggett)
- Newsgroups: rec.arts.int-fiction
- Subject: Re: Parser heresy (was Re: Searching for a sense of wonder)
- Date: 17 Nov 1992 07:30:06 GMT
- Organization: MIT Artificial Intelligence Laboratory
- Lines: 54
- Message-ID: <1ea71uINNkv1@life.ai.mit.edu>
- References: <1e5mvtINNnlp@terminator.rs.itd.umich.edu> <1992Nov15.155951.3262@starbase.trincoll.edu> <BxttrL.E6x@acsu.buffalo.edu>
- NNTP-Posting-Host: xbar.ai.mit.edu
-
- In article <BxttrL.E6x@acsu.buffalo.edu> goetz@acsu.buffalo.edu (Phil Goetz) writes:
- >Well, let me throw a bit of heresy at you which I believe: Parsing is not
- >a problem for adventure programs.
-
- Heresy? Well add me to the list of heretics. *Syntax* parsing is
- almost solved. We have principles-and-parameters parsers here that do
- extremely well and can parse up feeds off the news wires in real time.
- (Other places have similar things too.) Of course, they have no clue
- what the sentences are *about*, but they can sure make pretty trees for
- you, and they can tell you what all the lexical categories are.
-
- No, it's not a parsing problem. And when you point out that you
- wouldn't *want* to use most complex sentence types in a game,
- you're dead on the mark. Purely for playability reasons, we
- would NOT want to add, say, adverbs to our parser. Can you *imagine*
- the following:
-
- *-----------------------------------------------------------------------------*
- >get the book from the troll
- The troll sees your lame attempt from a mile away. He laughs in your face.
-
- >get the book from the troll very quickly
- You snatch the book away with a lightning-fast jerk.
-
- *** Your score just went up ***
- *-----------------------------------------------------------------------------*
-
- I have one thing to say: "NOT!"
-
- >2. Conversation. This is not a parsing problem, it is a knowledge
- >representation and reasoning problem. If you give me a good representation
- >for the concepts in a sentence, I can write a parser for it quickly.
- >Syntactic structure is easy.
-
- Again, right on the mark. Syntax is really hard, but it (again) is
- pretty well-understood. But *reasoning*, that's not undertood at all!
- Do you think Deep Thought plays chess the way Kasparov does? Not a
- chance. Most AI to date has consisted of clever programs, IMHO, and
- not anything like reasoning, or even a lame approximation thereof. It
- is no coincidence that every major upset in the computer chess world
- has been accompanied by a significant hardware advance.
-
- My personal view is that we should be working more towards
- understanding the fundamental elements of intelligence in animals, and
- then see how we could go about applying this knowledge in machine
- settings. Until that happens, we are left with, as Phil said, a VAST
- language *understanding* problem, but a minimal (syntactic) parsing
- problem.
-
- Dave Baggett
- --
- dmb@ai.mit.edu MIT Artificial Intelligence Laboratory
- ADVENTIONS: interactive fiction (text adventures) for the 90's!
- dmb@ai.mit.edu *** Compu$erve: 76440,2671 *** GEnie: ADVENTIONS
-