PIL OWL Ontology Meeting 2011-10-24

From Provenance WG Wiki
Revision as of 15:59, 31 October 2011 by Tlebo (Talk | contribs)

(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

Meeting Information

prov-wg - Modeling Task Force - OWL group telecon

Attendees

Attendees:

  • Jim
  • Daniel
  • Paolo
  • Satya
  • Stian
  • Khalid
  • Stephan

Discussions

owl:Time

We believe this is already covered by:

http://dvcs.w3.org/hg/prov/raw-file/tip/ontology/ProvenanceFormalModel.html#wasGeneratedAt
http://dvcs.w3.org/hg/prov/raw-file/tip/ontology/ProvenanceFormalModel.html#assumedRoleAt
http://dvcs.w3.org/hg/prov/raw-file/tip/ontology/ProvenanceFormalModel.html#startedAt
http://dvcs.w3.org/hg/prov/raw-file/tip/ontology/ProvenanceFormalModel.html#endedAt

Daniel, Khalid, Stian: discussed about use of time through entityInRole

Luc has raised point that time is not associated directly with entities, but through process executions, generation or use.

Paolo: An event line points out when events in time occur (?)

4 types of events: generation, use, beginning end of process execution

Satya: Don't believe it's correct that time is not associated with time.

Stian: Perhaps the problem is hasTemporalValue without using any of the 4 subproperties is a bit too ambigious and don't mean anything in PROV-DM.

Khalid: We need time for EntityInRole to say when an entity was used. We can allow association with EntityInRole without needing to associate it with Entity. With EntityInRole you can say when it was generated, in which role.

(Stian: If we use EntityInRole to do both generation and use times and roles, do we still include prov:wasGeneratedAt?)

All: Believe this is not an issue anymore.

Khalid: What if entity is playing two roles in the same PE. Then we would need 2 different entityInRole. (Stian: What about both EntityInRole being wasGeneratedBy the same PE?) An Entity can only be generated once by a process (Daniel). So this would simply be "invalid" or "inconsistent" just like making a loop of wasDerivedFrom. Yes (Daniel)

EntityinRole with example in PROV-O: crime file scenario, better description. Need a way to show same entity in two different roles to make this .

Subclassing of Agent is also used in examples - is this comparable to assuming a role?

Satya: Subclassing is specialisation, but roles are not specialisation.

Daniel: Reza B'Far raised the question to also have a non-human Agent example.

Khalid: With one entities playing two different roles in two different PEs.. could be difficult?

Satya: yes, two new entities, but linked to original with prov:wasAssumedBy

Is Location only geographical?

Daniel: Location is confusing. Geographical place, but in example we point to a path on a disk.

Satya: In our concept of Location, Geospatial is geographical space location. (??) Have put a note to clarify this.

Paolo:W hat is OGC?

Open Geospatial Consortium, Related to ISO? OGC works on standards for geospatial information. ISO-19115:2003, ISO-19115:2009, ISO-19157 19115:2003 - Greographic Information - Metadata

http://www.iso.org/iso/catalogue_detail.htm?csnumber=26020

19115-2:2009 - Geographic Information - Metadata - Part 2: Extensions for imagery and girdded data

http://www.iso.org/iso/catalogue_detail.htm?csnumber=39229

19157 - Geographic Information - Data Quality

http://www.iso.org/iso/iso_catalogue/catalogue_tc/catalogue_detail.htm?csnumber=32575

ISO Lineage basic "primer"- https://geo-ide.noaa.gov/wiki/index.php?title=ISO_Lineage

ISO 19115-2 basic info - https://geo-ide.noaa.gov/wiki/index.php?title=ISO_19115-2_(Geographic_Information_–_Metadata_Part_2)

Extension PROV-DM

Stian: attributed, etc. are already in OWL, needs to be added to HTML.

@Stian: you mean Paul's shortcuts?

@DGarijo - yes, section 7.2 . http://www.w3.org/TR/2011/WD-prov-dm-20111018/#common-relations

Great! I didn't see them yesterday.

  • wasQuotedOf
  • wasAttributedBy
  • wasSummaryOf

Satya: Should look at separating out (different OWL/HTML) specialisations at a later stage

Daniel: We are missing examples for:

-wasEventuallyDerivedFrom
-dependedOn
-hadParticipant
-wasComplementOf
-hadRecipe
-wasInformedBy
-wasScheduledAfter
-followed
-hasTemporalValue

Satya: What about cases where these are not covered by crime file example in PROV-DM?

The following concepts do not appear in the data model:

-followed
-hadRecipe is recipeLink (but "recipeLink" is not a noun in past tense) I know :(
-hasTemporalValue
-wasAssumedBy

Collection extension

Stian looked at Collection extension, and have a commit ready (on the wrong linux partition!) that adds this to OWL and HTML = but renames them from confusing wasAddedTo_Collection etc to perhaps more understandable subproperties of prov:wasDerivedFrom using expansion/reduction

(after discussions with colleagues who got very confused by PROV-DM terminology on this) :

(From memory:)

[a prov:Collection ] prov:expanded [ a prov:Collection ]
[a prov:Collection ] prov:wasExpandedBy [a prov:Entity] #value
[a prov:Collection ] prov:wasExpandedAt [a prov:Entity] #key
[a prov:Collection ] prov:reduced [ a prov:Collection ]
[a prov:Collection ] prov:wasReducedBy [ a prov:Entity]
[a prov:Collection ] prov:wasReducedAt [ a prov:Entity] # key
[ a prov:EmptyCollection ] # prov-DM does not have this, but I can find this very useful

Example of an ordered list:

:col0 a prov:EmptyCollection .
:col1 prov:expanded :col0 ;
prov:wasExpandedBy :e1 ;
prov:wasExpandedAt [ rdf:value 0 ] .
:col2 prov:expanded :col1 ;
prov:wasExpandedBy :e2 ;
prov:wasExpandedAt [ rdf:value 1 ] .
 
:col2 a prov:Collection, rdf:List ;
 rdf:value ( :e1, :e2 ) .
# rdf:List expansion here shown to indicate one particular interpretation of
# the abstract prov:Collection - its interpretation depends on the specialisation
# and types of keys.

Add ttl serialization

Daniel: We've used TTL in all emails because it's easier to read - so should we not also do that in the HTML? Satya: Can do something similar to how the OWL 2 primer with different buttons where you can choose how to do it.

TO DO: Tim/Daniel: Explore use of ttl serialization

Transitivity

Daniel: Have discussed some transitivity issues earlier. but HTML/OWL is sometimes out of date. Khalid: Will look at this. Satya: also Need diagrams for the shortcuts. Satya will add the HTML sections. ACTION Satya: HTML sections for 7.2 extensions in PROV-DM ACTION Khalid: Diagrams for 7.2 extensions in PROV-DM (already in OWL)

Outline of document vs PROV-DM

Khalid: Luc was also worried about the outline of the document not mapping straight to the PROV-DM outline. Satya: Don't think its neccessary for first working draft.