Thank you all for feedback and thoughts.
I find the comments on linguistics and speech acts interesting – I think it’s the right thread to pull on to find an answer to my original query. I don’t think
the discussion has yielded a clear position of “commands” within ontology yet, so here are a few more thought-provoking questions.
Using William Frank’s example, the proposition: “cat in a box” can easily be represented in an ontology. Representing: “Tom: put the cat in the box” (call
this command A) in an ontology involves that proposition coupled with a command “put”, a directing agent (me?), and a directed agent (Tom). I agree that A is a speech act (imperative, performative), but what is it in an ontology? Are we saying that
“speech act” is an object in an ontology, and
that “command” is a kind of “speech act”, and
the compound proposition A is an instance of “command”?
If that command is issued by one software agent to another (e.g., in a SOA-based workflow), is it still a “speech act”?
William C. Burkett
Booz | Allen | Hamilton
121 S Tejon St # 900 |
T: 719-387-6452 | M: 310-318-5500 | F: 719-387-2020
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx]
On Behalf Of Obrst, Leo J.
Sent: Saturday, April 28, 2012 11:04 AM
Subject: Re: [ontolog-forum] Ontology of Commands
Just a follow-up. You might look at Jose Vidal’s “Agent Communication” briefing:
FIPA (circa 2000) performatives
http://jmvidal.cse.sc.edu/talks/agentcommunication/performatives.html. There is also a pointer to
FIPA Communicative Act Library Specification.
This has been an interesting discussion, with lots of good points made. One observation that I will add is that the concept of command
(and SOAs) implies a relationship (part of the command context others have mentioned) between whatever agent(s) is issuing the command (or speech act) and whatever agent(s)/service provider(s) the command is directed at. That relationship constrains the scope
of the command, and determines whatever consequence management might be entailed by failure to perform whatever actions are being commanded, including any performance shortfalls (timeliness, completeness, or other quality attributes specified/expected by the
command). That is the major reason for SLA’s in SOAs. It’s also why one of the Net Centric Principles is (explicit and dynamic) relationship management. Note also that relationships have many operative scope dimensions, such as jurisdictional environment
(what laws and legal framework(s) apply to the relationship), business models, institutional context(s), if any, social environment (something commanded might be legal but taboo), and operational context(s) in which the relationship and associated command
might be applicable/valid.
On Behalf Of William Frank
Sent: Friday, April 27, 2012 8:41 PM
Subject: Re: [ontolog-forum] Ontology of Commands
>From what Leo is saying,
I take the underlying point to be, and it is a critical yet subtle one,
in model theory terms:
the ***model*** is what is really important: the world we are modelling, where things happen, (events), things persist (whatever you want to call entities or the like), etc. We've *got* to have one of these models, to think with.
A speech act is a kind of event, some actor performs the act. The propositional content of the act is not part of the type of the event. A request is an action-- an act in which the requestor asks another actor to do something (this something is not a subtype
of request), which other actor has the right to decide whether or not to do, based on the request. A command is a similar but different action -- an act in which the commander indicates an action that the commanded must perform, -- the commanded actor is
obligated to do as commanded.
The language semantics are the linquistic entities **refering** to the model. this has things like verbs in it, that generally **refer to** events, and nouns, which might refer to things that we model as events or happenings (perhaps avalanches) or to things
we model as persisting (cities). For example, the particle 'ma' in chinese, the interrogative symbol '?' the interogative facial _expression_ of raising the eyebrows. All different vocabulary items that refer to the same model element the event of asking.
We might or might not have one or more of these language sematics, separate from the model.
I hope this helps explain the essence of this, in an informal way.
On Fri, Apr 27, 2012 at 8:15 PM, Obrst, Leo J. <lobrst@xxxxxxxxx> wrote:
Yes, nouns and adjectives mostly, besides verbs. I think you are better to consider “events”. Almost all verbs, but quite a few nouns and adjectives do denote events. However, those
nouns and adjectives are typically based on the verbs. Destroy: destruction : destroyed. They all typically map the same natural language arguments into the verb-based predicates and then to the event denotation.
On Apr 27, 2012, at 8:00 PM, Obrst, Leo J. wrote:
Some (we do) model both the linguistic semantics side (e.g., verbs, etc.)
Does "etc" include nouns?
What's the ratio between verbs/commands vs nouns? Many more nouns than verbs I believe.
Who/what/how controls the nouns?
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
This email is confidential and proprietary, intended for its addressees only.
It may not be distributed to non-addressees, nor its contents divulged,
without the permission of the sender.