uima message
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]
Subject: Re: [uima] not Entity but ...
- From: David Ferrucci <ferrucci@us.ibm.com>
- To: Karin Verspoor <Karin.Verspoor@UCHSC.edu>
- Date: Wed, 14 May 2008 16:05:57 -0400
We intend for "entity" to
represent any thing in some domain of discourse that an annotation
may refer to. Including objects, relations, events, sets etc.
We consulted a few sources and chatted
with Chris Welty a bit on this. I think we may be OK with "entity".....
While it usage may have evolved in some
circles to exclude things like relations, it seems inclusive from the basic
philosophical and ontological perspective.
=====
An entity is something that has a distinct,
separate existence, though it need not be a material existence. In particular,
abstractions and legal fictions are usually regarded as entities. An entity
could be viewed as a set containing subsets. -- Wikipedia
"To be assumed as an entity is,
purely and simply, to be reckoned as the value of a variable." --
Quine
=====
A concern is that in NLP circles the
distinction seems common where name-entity detection is considered distinct
from relationship detection for example where arguments to relations are
"entities".
It is also common in database modeling.
Entity-Relationship models (Chen) for example make a clear distinction
between constructs that represent entities in the domain and constructs
that represent domain relations. Of course both these simply become relations
(i.e. tables) at the data model level.
I don't think 'denotation' is right.
Going back to are original intent this thing (now called 'entity') should
represent really anything that may be referred to including individuals.
I like "Referent" (adam's
suggestion) since this divorces us from any ontological commitment and
focuses on the role a domain object plays relative to an annotation.
That is, the annotation refers to it
and it is referred to by the annotation. Then any domain model with any
root may be built independently of 'Referent" and any concept in the
domain can play the role of a Referent with respect to an Annotation. This
approach while loosely coupling the domain model to the annotation structure,
introduces another layer of indirection through Referent. So any
additional referent object is required for ever domain object linked
to an annotation -- not good.
I think semanticType is wrong because
that is not the root of all domain objects. We may all agree a Person
is an Animal but I personally do not consider myself a 'semanticType' or
a 'denotation' for that matter. I do consider myself an Entity however.
And I would accept that I may be referred to by an annotation over
a mention in a document.
If we go back to Quine, "Entity"
is the root of the most basic ontology and subsumes objects, relations,
sets and even syntactic constructs like Classes and Predicates (this according
to Welty -- I did not cross check with Quine).
------------------------------------------------------------------------
David A. Ferrucci, PhD
Senior Manager, Semantic Analysis & Integration
Chief Architect, UIMA
IBM T.J. Watson Research Center
19 Skyline Drive, Hawthorne, NY 10532
Tel: 914-784-7847, 8/863-7847
ferrucci@us.ibm.com
------------------------------------------------------------------------
http://www.ibm.com/research/uima
Karin Verspoor <Karin.Verspoor@UCHSC.edu>
05/12/2008 02:34 PM
|
To
| uima@lists.oasis-open.org
|
cc
|
|
Subject
| [uima] not Entity but ... |
|
(possible options for changing the name of the Entity
type to be more general)
Denotation
Extension
Semantics
Concept
SemanticClass
SemanticType
I do like the suggestion of Denotation. But then
I'm not a philosopher so I'm not entirely sure the philosophers would
agree that this is an appropriate use of that term.
Karin
--
Karin Verspoor, PhD
Research Assistant Professor
Center for Computational Pharmacology, University of Colorado Denver
PO Box 6511, MS 8303, Aurora, CO 80045 USA
karin.verspoor@uchsc.edu
/ tel: (720) 279-4875 / campus: 4-3758
[Date Prev]
| [Thread Prev]
| [Thread Next]
| [Date Next]
--
[Date Index]
| [Thread Index]
| [List Home]