IW Meeting 2011-08-04

From Inference Web

Jump to: navigation, search

Contents

Meeting Information

Attendees

  • Paulo
  • Tim
  • Cynthia

Agenda

  • Cynthia IW Browser status
  • Paulo's derivation concerns

Discussions

NYT image retrieval and crop example

posted to inference-web.org:

IW,

Based on some discussions that Jim, Cynthia, Deborah, and I have had in the IW telecon, I created a new example to illustrate my current provenance capture encoding.

Please find the example at 

https://github.com/timrdf/csv2rdf4lod-automation/blob/master/doc/examples/source/nytimes-com/2011-07-22-world-europe-scene-of-the-norway-attack/version/2011-Jul-24/source/oslo3.png.pml.ttl
(you can get the actual file by clicking "raw", or
https://raw.github.com/timrdf/csv2rdf4lod-automation/master/doc/examples/source/nytimes-com/2011-07-22-world-europe-scene-of-the-norway-attack/version/2011-Jul-24/source/oslo3.png.pml.ttl

~~~~
Double bonus points for anyone that uses pcurl.sh to capture provenance while retrieving this provenance!
~~~~

There is subsequent provenance at
https://github.com/timrdf/csv2rdf4lod-automation/blob/master/doc/examples/source/nytimes-com/2011-07-22-world-europe-scene-of-the-norway-attack/version/2011-Jul-24/manual/oslo3.png.pml.ttl
(clicking "raw" gets you to
https://raw.github.com/timrdf/csv2rdf4lod-automation/master/doc/examples/source/nytimes-com/2011-07-22-world-europe-scene-of-the-norway-attack/version/2011-Jul-24/manual/oslo3.png.pml.ttl

Questions, discussion, and feedback most welcome.

Regards,
Tim

Some context for this conversation? Enabling the PML encoding in formats other than RDF

content negotiation and other serializations.

http://download.librdf.org/source/raptor2-1.9.1.tar.gz

rapper -g -c my.ttl  # This gives the count of triples, guessing the input syntax with -g question:  what are the possible input syntaxes that rapper recognizes?

  -i FORMAT, --input FORMAT   Set the input format/parser to one of:     rdfxml          RDF/XML (default)     ntriples        N-Triples     turtle          Turtle Terse RDF Triple Language     trig            TriG - Turtle with Named Graphs     rss-tag-soup    RSS Tag Soup     grddl           Gleaning Resource Descriptions from Dialects of Languages     guess           Pick the parser to use using content type and URI     rdfa            RDF/A via librdfa     nquads          N-Quads

question: who is behind rapper? Dave Beckett http://librdf.org/INSTALL.html

rapper -i turtle -c my.ttl # This gives the count of triples, forcing the parser to assume turtle. rapper -g -o rdfxml my.ttl > my.ttl.rdf rapper -g -o rdfxml-abbrev my.ttl > my.ttl.rdf

CI-Server += csv2rdf4lod

Tim and Aida need to touch base to better understand potential connection between Tim's CSV2RDF code and Aida's CI-Server

CI-Server relies on Drupal's content type, which may inform that a given file is of type CSV

PIL's ISSUE 26

Context: 1. P uses A  2. P uses B 3. P generates C 4. C derived from A

If the provenance of C is the concern of a user of C (as opposed to the  provenance of a process that generates C), one may have the following  questions:

1) What are the “uses” and “generates” relationships adding to one’s understanding of C if something is wrong with C?

2) Can we infer that __C was derived by__ the execution of process P? How?

some fundamental issues:

Why do we care about process P if nowhere is stated that P was responsible for the derivation of C? For instance, P may be just the execution of a straight database lookup that is "generating" C as a response to a database query but not necessarily "deriving" C  by some transformation from A (as stated in 4).

Why we care about 1 and 2 if we do not know exactly the role of P deriving A?

  • A and B can be roled by P (so can C)

- lets follow this rationalle:   1b. P uses A to be an input that will be used to derive C   2b. P uses B as an optimization argument that does not affect the derivation of C at all   3b. P generates C as a derivation   If that is the case, why we bother about modeling 2b if we only care about 4?  Why we cannot restrict "uses" and "generates" to roles of being part of a derivation? in this case, we would not need to model 2b and 4 would be derived from 1b and 3b.    Tim is hoping that pil:Roles become roled involvements (where the involvement designations fall outside of pil)

mybaking pil:uses :spoon_1  # If the spoon was not REQUIRED to lead to :yummy, then Paulo is concerned that we have more than we needed.
mybaking pil:uses :butter_23
mybaking pil:uses :joe, :sam,  :george
mybaking pil:generates :yummy

[:mybaking, :spoon_1, my:Mixer] [:mybaking, :butter_23, my:Ingredient] [:mybaking, :joe, my:Implementor] [:mybaking, :sam, my:RecipeReciter] [:mybaking, :george, my:KitchenHost]

The question for me regarding the roles in your cake example above is thde following: can you derive the :yummy without using :spoon_1? If so, the cake is not derived from spoon_1 and the fact that spoon_1 is irrelevant for the derivation of :yummy. 

Tim's question: What Paulo means by "derive"? Paulo's answer: A derived from B =~= A depends on B to exist (that is a possible point of view)

the yummy2 made by using my hand as a my:Mixer could be __the same or different__ as the yummy made by mixing with a spoon. (FRBR and notions of different)

From 4, I still do not have a clue how C was derived

TODO: Tim to email Aida the example from CI-Server.

References:

Personal tools
Navigation