🚀 go-pugleaf

Date: Thu, 4 Jun 87 13:38:38+0300
From: nsc!nsta!nsta.UUCP!iddo (Iddo Carmon /NSTA (052)-522-267)
Organization: National Semiconductor (Israel) Ltd.

>From: mlwh@sphinx (Martin Hall)
>I would be interested in finding out about document analysis.  I mean
>this in as general as a sense as you want to take it.  Any pointers
>would be appreciated.

My view is that this kind of activity is best handled by a proper mix of
human/machine interaction.  Consider the news system as an example: here
you have a massive amount of information to choose from, but still you're
able to handle it efficiently and select things that are relevant to you
by means of software utilities to various degrees of sophistication.

However, these utilities all rely on a set of conventions for putting
things in header lines that later enable the system to locate articles in
the newsgroup hierarchy, and on the intelligence of a human poster who
selects the proper newsgroups.  Also the structure of the newsgroup
hiereachy is developed by humans according to their interests and is a key
factor in the ease of selecting specific information.

Instead of treating a document as a 1-dimensional stream of characters and
trying to extract meaning from that, I'd like to see some common general-
purpose high-level 'document-programming' language evolving, that will be
used to annotate the text and will then enable automatic parsing of the
document into sections, threads of reasoning, selection of pieces by going
down a subject menue-tree, etc.  Such a convention may make it possible to
scan/archive documsnts according to their contents in numerous ways,
without a prerequisite for a "natural language understanding superexpert
system".

--
Iddo Carmon
Architecture Dept.                        Tel:  +972-52-522-267
National Semiconductor (Israel) Ltd.      uucp: ...!nsc!nsta!iddo
P.O.B. 3007, Herzlia B. 46104, Israel           {hplabs,pyramid,sun,decwrl}
----------------------------------------
Submissions to:   desktop%plaid@sun.com -OR- sun!plaid!desktop
Administrivia to: desktop-request%plaid@sun.com -OR-
sun!plaid!desktop-request
Paths:  {ihnp4,decwrl,hplabs,seismo,ucbvax}!sun

Chuq Von Rospach	chuq@sun.COM		Delphi: CHUQ

Now, where did my ex-wife put my Fairy Dust?