domain of gcis:relatedProject is the intersection of publication and dataset

[zednis]

gcis:relatedProject a owl:ObjectProperty ;
    rdfs:label "related project" ;
    rdfs:comment "A publication or dataset may be the result of one or more projects." ;
    rdfs:domain gcis:Publication, gcis:Dataset ;
    rdfs:range gcis:Project ;
    rdfs:subPropertyOf prov:wasAttributedTo .

I believe this domain would be the intersection of gcis:Publication and gcis:Dataset rather than the union.

I think we should generalize the domain.

[justgo129]

Sounds logical to me but I'll defer to @rewolfe and @bduggan on this matter.

[zednis]

based on the discussion of in USGCRP/gcis/pull/235.

I suggest we change the domain to prov:Entity and update the comment on the property to be "A project the entity is attributed to."

gcis:relatedProject a owl:ObjectProperty ;
    rdfs:label "related project" ;
    rdfs:comment "A project the entity is attributed to." ;
    rdfs:domain prov:Entity;
    rdfs:range gcis:Project ;
    rdfs:subPropertyOf prov:wasAttributedTo .

[justgo129]

Fine here aside from the grammatical issue in the definition.

[justgo129]

Also, to what aspect of the Project is the entity attributed? The existence, creation, etc?

[zednis]

Is there a difference between ascribing an entity's existence vs it's creation?

Definition of PROV Attribution:

Attribution is the ascribing of an entity to an agent. When an entity e is attributed to agent ag, entity e was generated by some unspecified activity that in turn was associated to agent ag. Thus, this relation is useful when the activity is not known, or irrelevant.

Roles can be used to with a qualified Attribution (see example at http://www.w3.org/TR/prov-o/#Attribution) to specify the role the agent had in the activity that generated the entity.

[rewolfe]

Shouldn't it be "wasRelatedTo", not just "relatedTo"?

Change "A project the entity is attributed to."
to "A project that was related to an entity."

I assume the use cases are:

a. Project A was related to Project B.
b. Project A was related to Organization B.

However, should we be more specific and use terms like "wasAffiliatedWith",
"wasFundedBy" and "wasPartOf"?

Examples:

c. Project A was affiliated with Project B.
d. Project A was funded by Organization B.
e. Project A was part of Project B.
f. Project A was part of Organization B.

On Wed, Sep 2, 2015 at 11:22 AM, Stephan Zednik notifications@github.com
wrote:

Is there a difference between ascribing an entity's existence vs it's
creation?

Definition of PROV Attribution:

Attribution is the ascribing of an entity to an agent. When an entity e is
attributed to agent ag, entity e was generated by some unspecified activity
that in turn was associated to agent ag. Thus, this relation is useful when
the activity is not known, or irrelevant.

Roles can be used to with a qualified Attribution (see example at
http://www.w3.org/TR/prov-o/#Attribution) to specify the role the agent
had in the activity that generated the entity.

—
Reply to this email directly or view it on GitHub
#143 (comment)
.

Robert Wolfe, NASA GSFC @ USGCRP, o: 202-419-3470, m: 301-257-6966

[zednis]

This property is used to assign attribution of some entity to a project. The super-property is prov:wasAttributedTo which has a specific meaning that holds between an entity and an agent and is more precise then a general 'related' property.

A general 'relatedTo' property would be a different property.

The two use cases @rewolfe provides are not appropriate a sub-property of prov:wasAttributedTo. Additionally the previous domain of the property - which was constructed erroneously as an intersection - indicates that the domain of this property was intended to be types like gcis:Publication or gcis:Dataset and not projects.

[justgo129]

I'll defer to @rewolfe going forward with this ticket.

[rewolfe]

Okay, I am sorry that I misunderstood the context. So, an example would be:

Dataset A was related to Project B.

I think this is pretty general, but it may be okay. More typical term
would be "was produced by".

So to relate a dataset to a project, we need two triples:

Dataset A was produced by Activity B.

Activity B was related to Project C.

Again, a more typical term way of relating an activity and project would be:

Activity B was performed under Project C.

or

Activity B was part of Project C.

I'm not sure if this helps.

On Sep 3, 2015 9:14 AM, "justgo129" notifications@github.com wrote:

I'll defer to @rewolfe https://github.com/rewolfe going forward with
this ticket.

—
Reply to this email directly or view it on GitHub
#143 (comment)
.

[zednis]

@rewolfe The current definition of gcis:relatedProject asserts that it is a sub-property of prov:wasAttributedTo.

This means that the statement
:Dataset_A prov:wasAttributedTo :Project_C
is inferable from
:Dataset_A gcis:relatedProject :Project_C

In PROV attribution has the following definition - Attribution is the ascribing of an entity to an agent. When an entity e is attributed to agent ag, entity e was generated by some unspecified activity that in turn was associated to agent ag. Thus, this relation is useful when the activity is not known, or irrelevant.

So in PROV attribution is a shorthand for saying:
:Dataset_A prov:wasGeneratedBy :Activity_B
:Activity_B prov:wasAssociatedWith :Project_C

This conforms to your suggested statements:

Dataset A was produced by Activity B.
Activity B was related to Project C.

[rewolfe]

@zednis Got it! Thanks for being patient in helping me understand this.

So, yes, I'm okay with the definition: "A project the entity is attributed
to."
However, would is be clearer?: "An entity ascribed to a project through an
unspecified activity."

On Fri, Sep 4, 2015 at 12:37 PM, Stephan Zednik notifications@github.com
wrote:

@rewolfe https://github.com/rewolfe The current definition of
gcis:relatedProject asserts that it is a sub-property of
prov:wasAttributedTo.

This means that the statement
:Dataset_A prov:wasAttributedTo :Project_C
is inferable from
:Dataset_A gcis:relatedProject :Project_C

In PROV attribution has the following definition - Attribution is the
ascribing of an entity to an agent. When an entity e is attributed to agent
ag, entity e was generated by some unspecified activity that in turn was
associated to agent ag. Thus, this relation is useful when the activity is
not known, or irrelevant.

So in PROV attribution is a shorthand for saying:
:Dataset_A prov:wasGeneratedBy :Activity_B
:Activity_B prov:wasAssociatedWith :Project_C

This conforms to your suggested statements:

Dataset A was produced by Activity B.
Activity B was related to Project C.

—
Reply to this email directly or view it on GitHub
#143 (comment)
.

Robert Wolfe, NASA GSFC @ USGCRP, o: 202-419-3470, m: 301-257-6966

[zednis]

I am ok with "An entity ascribed to a project through an unspecified activity" if everyone else is.

I was trying to follow the grammatical structure that dublin core terms uses in their comments, but I don't think it works well in this case.

[justgo129]

👍

[rewolfe]

+1

On Fri, Sep 4, 2015 at 2:08 PM, justgo129 notifications@github.com wrote:

[image: 👍]

—
Reply to this email directly or view it on GitHub
#143 (comment)
.

Robert Wolfe, NASA GSFC @ USGCRP, o: 202-419-3470, m: 301-257-6966

[zednis]

ok, updated proposal with the new comment text:

gcis:relatedProject a owl:ObjectProperty ;
    rdfs:label "related project" ;
    rdfs:comment "An entity ascribed to a project through an unspecified activity" ;
    rdfs:domain prov:Entity;
    rdfs:range gcis:Project ;
    rdfs:subPropertyOf prov:wasAttributedTo .

Since I think we already have consensus I will wait a bit and baring any objections I will submit the pull request.

[xgmachina]

+1

[justgo129]

Closed #143 due to merged #147.

[justgo129]

@zednis @xgmachina now that #147 has been merged, please feel free to proceed editing the turtle templates where necessary. It may also impact #17.