Re: Ontolingua to SQL

Tom Gruber <gruber@HPP.Stanford.EDU>
Message-id: <199405160256.TAA20752@HPP.Stanford.EDU>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Sun, 15 May 1994 19:58:36 -0800
To: (Ivan), ontolingua@HPP.Stanford.EDU
From: Tom Gruber <gruber@HPP.Stanford.EDU>
Subject: Re: Ontolingua to SQL
In-reply-to: <>

At  1:25 PM 5/4/94 +0010, Ivan wrote:
>I am currently working on a project that investigates the mapping
>of Ontolingua into SQL.
>We are identifying a suitable representation ontology (Probably a subset
>of the Frame Ontology) which allow us to establish the basis for our
>We aim at coming up with a set of guidelines to turn ontologies expressed
>in Ontolingua into the conceptual schema of a relational database. Likewise,
>we also aim at mapping KIF into SQL (Query language). This allows us to have
>a generic KIF interface for assertion and query to the database.
>I would not be surprised if somebody has already done some work on this
>area. If this is the case, I would be very grateful if anyone could point
>me towards some relevant bibliography or some existing results.

We don't have anybody building a KIF->DB schema or SQL tool, but that's
definitely one of the applications we would like to see and think would pay
off.  For those of you who follow the standards business, the ANSI
conceptual schema committee (called IRDS) just recently issued a report on
this topic that suggests that the way to support interoperability among
information resources is to provide a semantic foundation rooted in logic.
So go for it!

A couple of small suggestions to get started.  The KIF output of ontolingua
is probably the closest thing to SQL.  So one strategy would be to generate
KIF and run that through an ad hoc translator into SQL or the schema
definition language.  If you give us some hints about a better canonical
form we can change the ol-to-KIF translator to suit.   Second suggestion:
use the bibliography ontology as an example test case.  That was explictly
designed as a test case for database integration.  Most of the definitions
are within the expressive power of database schema languages and the target
domain is, in fact, bibliographic information of the sort stored in endnote
and similar database products.