IW Meeting 2010-12-15

From Inference Web

Jump to: navigation, search

Contents

Meeting info


(everyone at AGU)

Attendees

  • Tim (skype in)
  • Deborah
  • Stephan
  • Paulo
  • Cynthia
  • Li (skype in)

Agenda

(suggested by Paulo)

1) Generic PML 1.1) Brainstorm an approach to handle multiple versions <--- we are HERE li had a ppt of flow. with examples of extensions for stephan and extensions for tim.


2) PML 2.1 2.1) Discuss proposed PML changes according to Inference Web wiki


3) New W3C FOO (provenance) language 3.1) Update everybody in terms of W3C Provenance Incubator activities including FOO 3.2) Brainstorm a strategy to upgrade PML tools to support FOO


(remaining suggestions)


Discussions

multiple versions

2.0 then release 2.01 for Stephan need to move to 2.1 e.g. 2.tim

simultaneous updates

Paulo is concerned about branching.

Deborah: branching is exploratory homework to get to next point release faster.

Stephan: variety of branches along with point releases of "official" PML point releases.

Problem: once published "homework branches", then what?

We should NOT have two versions that use the same namespace (e.g. 2.01 reusing 2.0)

Stephan: How does Cynthia's tools "recognize" 2.01/2.tim's NodeSet just like 2.0's NodeSet? ????? - what was resolution?

Paulo: PML is probably NOT the first ontology to have this problem.

Stephan: Should our tools just use approved version?

Paulo: We are trying to solve a problem that we can still survive without solving the problem.

Deborah: e.g. Stephan needs extensions and can't wait for IW PML group to catch up b/c slow to produce new version. Will always be slower than any project wants us to be.

Stephan: everything in new namespace, operationally their not same. Cynthia: Should have generic behavior? Stephan: changing current PML instances to namespace pml.sc - no tool will load it.

Paulo: documents into triple store and can query. Another valid and consistent document added, breaks whole thing. violating constraints (only one timestamp). Invidually consistent, but union breaks b/c of cardinality. Was using pellet and TDB. ALL of this is with single version of ontology.

Paulo: multiple versions is a dream land.

Deborah: cardinality and range restrictions have undergone relaxation throughout the history of PML development.

Stephan: multiple layers: classes only, then add properties, then restrictions. (credited Tim, thinks it is Eric's ontology decomposition work)

Using deeper layers gives inferencing, using just top layers lets you know what properties are there.

Deborah: all domains and ranges GONE (please)

Stephan: PML Lite (v stephan) with no restrictions.

Deborah: different granularities of timestamps - getting rid of cardinality restrictions makes sense.

Deborah/Stephan: PML validation - informational warnings. problem is valid PML won't load into tools.

Stephan: validation service outside of PML ontology to help.

different: VALIDATION SERVICE and TIERED ONTOLOGY

Stephan: Visualization tools should not be using restrictions

Deborah: But they may be needed to make better visualizations.

Tier 1: no restrictions Tier 2: add restrictions

Cynthia: IWbrowser is general: looks for nodeset (and few others). Will show anything you've added into anything.

Stephan: PML Lite: without restrictions, domains, ranges

Deborah: We should get a proposal for a tiered

http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.87.6791&rep=rep1&type=pdf

W3C FOO

Tiered PML might be the solution for adopting FOO.

Most of PML concepts are in FOO, less of OPM's are in.

When we adopt handling FOO, our major task would be to rename concepts in PML ontology.

W3C will suggest concepts and constraints.

Paulo: what is truly challenging is [reconciling with FOO?]

Paulo: the more constriaint relaxation we do now, the more compatible we will be with FOO.

Deborah: using technology x, if I don't have rdfs:domain, can't do x b/c y wanted to know domain and range to use their tool (e.g. to use mysql)

Paulo: Person author of publication.


versioned

Paulo: does layered solve versioned?

PML Core shouldn't change frequently.

Core conflicts with W3C

PML Base not good b/c of IW Base.

Calling it PML Lite for today.

Paulo wants to get away from multiple versions. But we're all going to extend PML Light that live a separate life.

Deborah: But we'll have everything compliant with PML Lite (maybe change every 2 years). Deborah: 15 extensions per year.

Paulo: 100s of new users of FOO; will be messy.

Paulo: OPM people have things done, compromising to minimum to leverage tools they have. We are doing same thing. We wont' have our tools or their tools, but FOO tools.

Deborah: PML Layers will have 2 layers. Stephan:

PML Foundation is new name for PML Lite/PML Core, etc.


Paulo: Foundataion vs pml-p pml-j? How to slice it?

Paulo: Convo with Luc of PML data vs pml metadata provenance data vs provenance metadata is "provenance metadata" == provenance of provenance? conclusion (with Paulo/Luc): no. "provenance metadata" is not provenance of provenance.


Stephan to take lead on PML Foundation, helped by Tim and Nick.


FOO

incubator group FOO starting point for provenance.


(Can hear Deborah, Paulo, Stephan CANNOT hear Cynthia)

Roles

http://inference-web.org/wiki/PML_Ontology_Update_Request_6:_Roles_for_Antecedents

Paulo: What is the inferenceRule?

BEFORE: provenance:infstep_1

 a pmlj:InferenceStep ;
 pmlj:hasInferenceEngine tw_converter:csv2rdf4lod_0552153e6e1fdac8e196bb565a0a9d82 ;
 pmlj:hasAntecedentList _:node15j0dskq0x46 
 
 

PAULO: provenance:infstep_1

 a pmlj:InferenceStep ;
 pmlj:hasInferenceEngine tw_converter:csv2rdf4lod_0552153e6e1fdac8e196bb565a0a9d82 ;
 pmlj:hasAntecedentList _:node15j0dskq0x46 ;
 pmlj:hasMetaBinding [ pmlj:mapFrom " http://logd.tw.rpi.edu/source/fludb-org/provenance_file/animal-surveillance/version/2010-Nov-30/manual/avian.txt.csv"; 
                                     pmlj:mapTo "http://inference-web.org/registry/ROLE/Input.owl#Input" ];
 pmlj:hasMetaBinding [ pmlj:mapFrom "http://logd.tw.rpi.edu/source/fludb-org/provenance_file/animal-surveillance/version/2010-Nov-30/manual/avian.txt.csv.e1.params.ttl"; 
                                     pmlj:mapTo "http://inference-web.org/registry/ROLE/Parameters.owl#Parameters" ];
                                     # DOING a MAPPING to NOTHING.

.


PAULO 2: provenance:infstep_1

 a pmlj:InferenceStep ;
 pmlj:hasInferenceEngine tw_converter:csv2rdf4lod_0552153e6e1fdac8e196bb565a0a9d82 ;
 pmlj:hasAntecedentList _:node15j0dskq0x46 ;
 pmlj:hasInferenceRule <http://inference-web.org/registry/DPR/GMP.owl#GMP>;
 pmlj:hasMetaBinding [ pmlj:mapFrom " http://logd.tw.rpi.edu/source/fludb-org/provenance_file/animal-surveillance/version/2010-Nov-30/manual/avian.txt.csv"; 
                                     pmlj:mapTo "A" ];
 pmlj:hasMetaBinding [ pmlj:mapFrom "http://logd.tw.rpi.edu/source/fludb-org/provenance_file/animal-surveillance/version/2010-Nov-30/manual/avian.txt.csv.e1.params.ttl"; 
                                     pmlj:mapTo "B" ];
                                     # mapTo's value maps to a variable defined in scope of the InferenceRule identified by this inference step.

.


Tim: BUT THEY'RE URIs in LITERALS, and they are not refering to the _actual_ antecedents :-(

Paulo: BUT nobody knows what it means b/c it is not declarative.

Cynthia is posting http://registrar.inference-web.org/iwregistrar/ViewProvenance?uri=http%3A%2F%2Finference-web.org%2Fregistry%2FDPR%2FGMP.owl%23GMP&mode=full&reqreg=yes

FOO

 I'm pleased to announce publication of:

   Provenance XG Final Report
   http://www.w3.org/2005/Incubator/prov/XGR-prov-20101214/


For the past year, the Provenance Incubator Group worked on use cases and
requirements for provenance on the Web, providing a synthesis of the
state-of-the art, and developing a roadmap in the area of provenance and
possible recommendations for standardization efforts.

In the report, the XG highlights the importance of provenance and presents
requirements in a variety of contexts in the Web based on use cases
collected from the community. Based on these requirements and a close
examination of existing provenance work, the XG identified the need for
standard mechanisms to represent and access provenance. The group
formulated a roadmap for provenance on the Web that includes short term
and long term priorities. The group also agreed to concrete starting
points to ensure rapid progress towards a standardization effort, and
proposed a charter for a Provenance Interchange Working Group.

Read more in their conclusion:
http://www.w3.org/2005/Incubator/socialweb/XGR-socialweb/Overview.html#Conclusions


Congratulations to the XG. With this publication, the Provenance Incubator
Group is now closed.

Coralie Mercier, Incubator Activity Lead

Actions

  • Tim to explore 2.tim copy vs 2.lite.tim tiered.
  • team to pursue layered 2.1
  • Stephan to propose PML Foundation, Tim to help, Nick to help.
Personal tools
Navigation