Re: A proposal for modeling agents

Hi stephan,

There was no request for a formal vote on this issue.
In general, there was support for the approach.
As editor of the document, We decided that the proposition needed to be written in full to understand its implication.  This is work in progress. We'll notify the WG when complete.
Normal issues can then be raised.

Professor Luc Moreau
Electronics and Computer Science
University of Southampton
Southampton SO17 1BJ
United Kingdom

On 21 Nov 2011, at 18:20, "Stephan Zednik" <zednis@rpi.edu<mailto:zednis@rpi.edu>> wrote:

Has the agent proposal been voted on?  In Issue-155 Luc states that wasControlledBy will be replaced by terms from Yolanda's proposal - this seems to infer that this proposal has been accepted.

I think there are still some issues with this proposal that need to be discussed

On Nov 10, 2011, at 10:32 AM, Yolanda Gil wrote:

1) An "agent" is a type of entity that takes an active role in an activity such that it can be assigned some degree of responsibility for the activity taking place.

2) Many agents can be "associatedWith" a given activity.

While there is nothing wrong with this statement, 'associatedWith' does not imply an entity with agency nor a relation to an activity.  It is too broad.

associated |əˈsōsēˌātid, -SHē-|adjective(of a person or thing) connected with something else: two associated events.
I think associatedWith is a poor replacement for wasControlledBy and hadParticipant.  It is too broad and the semantics are not useful.  I suspect that every relation defined in prov-dm could be viewed as a valid specialization of 'associatedWith'.

wasDerivedFrom, wasQuoteOf, wasGeneratedFrom, etc. are all valid examples of one thing associatedWith another thing.

The current wasControlledBy and hadParticipant relations have useful semantics.  They imply agency by the involved entity (therefore an agent), and the existence of a relation between the agent and some activity, and are clearly distinct from used and wasGeneratedFrom.


3) Subclasses of agent are "foaf:person", "foaf:organization", and "software agent".

4) Agents can run activities on behalf of other agents, indicated by "runOnBehalfOf".

As long as Agents are a specialization of Entity and not explicitly defined in the context of an association/involvement/participation with an activity, it does not make sense to say that one Agent runOnBehalfOf Agent - we lose connection to the action being performed. If an Agent was used to describe me, and in an activity I act on behalf of RPI, my agent will now always be acting on behalf of RPI.  If my agent description is re-used in the context of other activities, we may be implying the involvement of a party that is not involved.

I think runOnBehalfOf must be defined within the context of a specific involvement between an agent and an activity - and we can already do that through qualifiers.

--Stephan



5) Agents can be responsible for starting and ending activities, indicated as "wasStartedBy" and "wasEndedBy".

Received on Monday, 21 November 2011 20:24:08 UTC