ontolog-forum
[Top] [All Lists]

Re: [ontolog-forum] Use of SUMO and "patient" of a Process

To: cassidy@xxxxxxxxx, ontolog-forum@xxxxxxxxxxxxxxxx
From: MDaconta@xxxxxxx
Date: Sat, 12 Jul 2003 17:04:30 EDT
Message-id: <133.228f5e54.2c41d1de@xxxxxxx>
Thanks Pat,

I have not done any work in NLP (so this is new).  But I will check
out the resource your mention.  You can see how I handled it so far
in the V0.1 protege ontology.

Thanks for the feedback,

- Mike

In a message dated 7/12/2003 1:46:40 PM US Mountain Standard Time, pcassidy@xxxxxxxxxxxxxxxx writes:

-- "patient" is common in NLP usage as one of the arguments
of transitive verbs (usually the "object" in an active-mood
sentence) -- one reference:
  http://oak.colorado.edu/~spradhan/publications/cuaq.pdf

They list the "semantic roles"  at one point:

"However, most of these semantic units are quite specific in what they
label. We identify a small set of thematic roles &shy; viz., agent,
patient, manner, degree, cause, result, location, temporal, force,
goal, path, percept, proposition, source, state, and topic, in the
candidate answer sentences, using a statistical classifier [5]. "




>>>2. the range of the property is "Entity" which includes both Physical and
>>>Abstract things.  Since the Process is a subclass of Physical, the
>>targets
>>>of the Process should probably have a range of "Object" (subclass of
>>>Physical).
>>
>>A patient of an event could be information, which isn't physical, even
>>though the event itself is situated in space and time, due to its
>>relationship to an Agent.
>
>
>
>Ok, I'll buy that.  Though Entity includes everything which is too broad.
>Some restraints on the target of a process are appropriate.
>

  Right.  The specific constraints will be identified in the
representations of specific types of process (or event).
The patient of "read" (in the usual sense) will be, for
example, a "Text-Physical", as in reading the creation date
on an Invoice.  Other senses of the same word (i.e. other
distinct though related identifiable processes)
would have different patients (e.g. to read the data off an
instrument).



---------------------------------------------------
Michael C. Daconta
Chief Scientist, APG, McDonald Bradley, Inc.
www.daconta.net
<Prev in Thread] Current Thread [Next in Thread>