Skip to end of metadata
Go to start of metadata

from Stian and Jits to Everyone:
Khalid: No agenda yet. We'll go through what people have been up to . We'll talk about the RO model work on the wiki (although it's not ready yet), and then..
from Stian and Jits to Everyone:
in the paper there were some questions/suggestions from Marco about preservation
from Stian and Jits to Everyone:
we'll wait just another minute
from Rafa(UPM) to Everyone:
hi everybody!
from Pique to Everyone:
hi rafa
from Jose to Everyone:
hi! I'm trying to get audio
from Stian and Jits to Everyone:
we can shout
from Raul to Everyone:
Hi !
from graham Klyne to Everyone:
Sorry I'm late... twice in two days (sad)
from Stian and Jits to Everyone:
Item 1. Roundtable
from Stian and Jits to Everyone:
Khalid: Stian/Jits/Sean/Khalid worked on the RO model, needed by architecture by 1st of sept. Still working, some issues to clarify/decide, but rougly the idea is..
from Stian and Jits to Everyone:
we'll use existing models. RO model, overall an RO is an aggregatio, annotations and some vocabulary
from Stian and Jits to Everyone:
for aggregation we'll use ORE
from Stian and Jits to Everyone:
we'll be using the annotation, and discussiong which model, Annotation Ontology (AO) Tim Clark, PAolo Cicerre or OAC
from Stian and Jits to Everyone:
both similar, based on Annotea
from Stian and Jits to Everyone:
we've got examples of both
from Stian and Jits to Everyone:
third one is vocabularies, this is where we're building now, where we're feeding up thing
from Stian and Jits to Everyone:
there's a wiki page where we're collecting this
from Stian and Jits to Everyone:
we'll send this out later for comments
from Stian and Jits to Everyone:
questions?
from graham Klyne to Everyone:
Is there a link to the current RO model doc?
from graham Klyne to Everyone:
Can't find it in the email.
from Jose to Everyone:
not anymore. I've been disconnected...
from Stian and Jits to Everyone:
we'll send the link once we're happy to go for more of a review phase - but it's there on the wiki
from Raul to Everyone: http://www.wf4ever-project.org/wiki/display/docs/RO+model
from graham Klyne to Everyone:
Ah, got it: http://www.wf4ever-project.org/wiki/display/docs/RO+model
from graham Klyne to Everyone:
I understand it's work-in-progress
from Stian and Jits to Everyone:
Graham: What are the differences between AO and OAC
from Stian and Jits to Everyone:
Khalid: Also other aspects to consider
from Pique to Everyone:
I've also been disconnected for 5 min.
from Stian and Jits to Everyone:
Marco and Kristina both here
from Stian and Jits to Everyone:
Jits: Talk to Marco first?
from Stian and Jits to Everyone:
Khalid: Wanted to clarify something Marco suggested with the paper
from Stian and Jits to Everyone:
on preservation
from Stian and Jits to Everyone:
when the service changed, when the data changed
from Stian and Jits to Everyone:
run the workflow on different data
from Stian and Jits to Everyone:
not clear to Khalid
from Stian and Jits to Everyone:
could you elaborate?
from Stian and Jits to Everyone:
Marco: wanted to differentiate reproducible execution, reproducible experiment
from Stian and Jits to Everyone:
affects definition of reproducible
from Stian and Jits to Everyone:
expectation management
from Stian and Jits to Everyone:
Marco: Would be hard to over many years guarantee a workflow to be reexecutable
from Stian and Jits to Everyone:
Khalid: understood this, but not clear to me in the use case, a workflow with a different user, changing the wf
from Stian and Jits to Everyone:
by adding service or change structure in order to build his own wf
from Stian and Jits to Everyone:
might implement differnet functionality from original wf
from graham Klyne to Everyone:
@stian: didn't we settle that "reproducible" is making it possible for someone else to run the w/f in a different environment and get comparable results?
from Stian and Jits to Everyone:
Marco: Is the wf synonymous to the method
from Stian and Jits to Everyone:
Khalid: So you might come up with a "completely different" wf, but achieving the same method
from Stian and Jits to Everyone:
Marco: the wf would be pretty similar. In one example, in my wf I query every day for the (question) iinformation
from Stian and Jits to Everyone:
some service deprecated by ddbj
from graham Klyne to Everyone:
Marco: "completely different" would be an extreme case...
from Pique to Everyone:
what makes a Wf different from another one ?
from Stian and Jits to Everyone:
to do this again I would have to replace the service with a different one with a different interface, functionally the wf would not do the same
from Pique to Everyone:
versioning
from Stian and Jits to Everyone:
Marco: always refer to the first version on myExperiment, even if it is disfunctional, because that's the one I initially published
from graham Klyne to Everyone:
@pique I think there are several answers to that: exact same bits, same fucxntion in different language, same result with different services, etc...
from Stian and Jits to Everyone:
Khalid: the new wf you define, would it belong to a new RO, or to the existing RO?
from Stian and Jits to Everyone:
Marco: On myExperiment you can refer to a wf with or without version number. It is a particular version that I have published. Would like to refer to the original one.
from Stian and Jits to Everyone:
Khalid: Would it still be interesting when the wf can't be executed any more, still to keep the RO as it is?
from Stian and Jits to Everyone:
Marco: Yes, exactly, useful and inevitable
from Stian and Jits to Everyone:
Khalid: Even broken wf on myExperiment still have value as you can see the design of the experiment
from graham Klyne to Everyone:
(I'm currently drafting some similar ideas for the Provenance WG - re. provenance of thing that changes)
from Stian and Jits to Everyone:
Marco: Interested in what Kristina and Pique thinks!
from Stian and Jits to Everyone:
Pique: Something of value is not just wfs that can be reproduced or results.. also the method
from Stian and Jits to Everyone:
having a look at the graph, you can see what the experiment is
from Raul to Everyone:
This is also part of the LO lifecycle question I raised some days ago
from Stian and Jits to Everyone:
great value is not just repeatabiliity, etc, even a broken wf, where you can see what you can achieve by assembling components
from Stian and Jits to Everyone:
still can get the idea
from Stian and Jits to Everyone:
and reuse that graph knowledge
from Jose to Everyone:
exactly! It is the knowledge about the method that we are really preserving
from Stian and Jits to Everyone:
Pique: interesting what MArco is saying, in preserving workflows, when you want repeatability or reprodubility, that's a minor issue for reviewing a paper when it's not years gone.. years later the knowledge is of higher value, just lookin at the picture/graph/workflow
from Stian and Jits to Everyone:
Jose: Knowledge of the method that we are preserving
from Stian and Jits to Everyone:
Khalid: Understand the example MArco was talking about now
from Stian and Jits to Everyone:
... (question) preserving the in such an extent this is less difficult than preserving the knowledge
from Stian and Jits to Everyone:
Jose: Many points here, preserving the execution, say in 10 years a piece of code can be executed again
from Stian and Jits to Everyone:
what we're focusing on is the knowledge and method
from Stian and Jits to Everyone:
the scientific method, to share it within a community and beyond
from Stian and Jits to Everyone:
the main thing is to understand the workflow as an understand of the method (question)
from graham Klyne to Everyone:
A user-case I'm looking at is using variations of a process (w/f) to improve quality of results. This needs preservation of w/f execution as a baseline.
from Stian and Jits to Everyone:
Khalid: Marco mentioned that when he specify a workflow, he specifies an abstract workflow, where the steps might not exists (service not known yet)
from Stian and Jits to Everyone:
capturing more the method than an executable method
from Stian and Jits to Everyone:
Marco: Could be the case, need to mention that the abstract wf are part of the living RO
from Stian and Jits to Everyone:
work in progress
from Jose to Everyone:
@Graham: Can you elaborate a little on that?
from Stian and Jits to Everyone:
may be capturing the method more than executable, bu tnot sure about every detail
from graham Klyne to Everyone:
@jose: yes, it's in an email I sent a couple of weeks ago.
from Stian and Jits to Everyone:
Khalid: Even as a living RO, sounds like the abstract workflow can be used as a kind of annotation of the final executable workflow
from Stian and Jits to Everyone:
keep it
from Stian and Jits to Everyone:
workflow annotated with names about what is meaningful
from Stian and Jits to Everyone:
abstract workflows can still be of value
from Stian and Jits to Everyone:
for annotating methods
from Jose to Everyone:
@graham: can you point me to that?
from Stian and Jits to Everyone:
Kristina and Marco agress
from Stian and Jits to Everyone:
Pique: Abstract workflows are of great value also because they can be useful for other communities
from Stian and Jits to Everyone:
the more abstract, the more useful across communities
from Jose to Everyone:
@pique: +1
from Stian and Jits to Everyone:
Kristina: Great value of a wf is that you have acommon vocabulary, input, output, etc.
from Stian and Jits to Everyone:
in an abstract workflow, it might help undersdtand a method, but not the same common vocabulary
from Stian and Jits to Everyone:
Marco: Importance of annotations
from Stian and Jits to Everyone:
annotation and metadata.. abstract/executable wf are still useless if everythign is called "workflow", (Stian: .. and "service" and "parameter", "arg1")
from Stian and Jits to Everyone:
that is why Marco finds the RO model important
from Stian and Jits to Everyone:
a framework for attaching all those annotations
from Jose to Everyone:
@Marco @Kristina: good point. How useful is a wf without their annotations?
from Stian and Jits to Everyone:
Pique: Versioning question
from graham Klyne to Everyone:
Some of this seems to be overlapping David's Scientific Social Objects discussion piece...
from Stian and Jits to Everyone:
in abstract workflows..
from Stian and Jits to Everyone:
how an abstract wf evolves to a more concrete.. when would it be a different wf?
from Stian and Jits to Everyone:
an issue to consider
from Stian and Jits to Everyone:
the evolution from abstract to living, particular wf
from Stian and Jits to Everyone:
a useful abstract wf can be reused in different directions
from Stian and Jits to Everyone:
Khalid: with specific workflow, do you mean executable wf?
from graham Klyne to Everyone:
@jose... looking still...
from Stian and Jits to Everyone:
Pique: (question) .. how to refer to wf's/authors, reusing a wf. How to refer to that workflow
from Stian and Jits to Everyone:
Khalid: Having an abstract workflow might give us the mean to say when a workflow is a version of another one, or a different workflow
from Stian and Jits to Everyone:
given two concrete workflows implementing same abstract wf, might be a sign/hint that one is a version of another
from Stian and Jits to Everyone:
Khalid: Two workflows implementing differnet abstract wfs, then they are not really considered version of each-other, evne if they have evolved from another
from Stian and Jits to Everyone:
Raul: related to question I raised on the list on life cycle
from Stian and Jits to Everyone:
Raul: How the abstract wf is related to the state of the RO
from Stian and Jits to Everyone:
abstract, living RO
from Stian and Jits to Everyone:
or the abstract RO can be part of a published RO
from Stian and Jits to Everyone:
s/RO/WF/
from graham Klyne to Everyone:
@Jose: 3 Aug 2011, "Use case and requirements for workflow preservation and information quality"
from Stian and Jits to Everyone:
Khalid: Yeah, did not get a final answer on that
from Stian and Jits to Everyone:
an abstract wf, if it exists, should be part of the RO, because it has value, with annotations it can give us additional information..
from Stian and Jits to Everyone:
we should not require that a RO must have an executable wf
from Jose to Everyone:
@graham: tx!
from Stian and Jits to Everyone:
s/RO/published RO/
from Stian and Jits to Everyone:
Jose: Looking at the RO model, can't see the workflow as an entity
from Stian and Jits to Everyone:
Khalid: In the RO model they would be a resource, which the RO is an aggregation of. No constraints so far, like "at least 1 workflow"
from Stian and Jits to Everyone:
Jose: With verisoning of workflows, and RO model, then they should explicity appear in the model
from Stian and Jits to Everyone:
@Jose - we've got workflow in the model as something you can annotate, but not yet ellaborated on the vocabulary. For example: https://github.com/wf4ever/ro/blob/master/examples/annotations-oac.trig#L186
from Jits to Everyone:
Marco: distinction between a method and a workflow
from Jits to Everyone:
don't mind abstracting away from a workflow
from graham Klyne to Everyone:
Sounds to me like a FRBR-style distinction
from Jose to Everyone:
@MArco: +1
from Stian and Jits to Everyone:
@Jose here it says that the :workflowProxy (which is the workflow as it is within the RO) is a rel:Workflow
from Stian and Jits to Everyone:
Marco: if you leaave it to the user to select what goes into a published RO from a living RO
from Stian and Jits to Everyone:
living could contain everything, the user selects what is part of the published/preserved RO - this could be different in different domains
from Stian and Jits to Everyone:
Marco: Users to annotate different things in RO as they want
from Stian and Jits to Everyone:
@Raul - mute keyboard please (smile)
from Stian and Jits to Everyone:
Marco: Put an interface on top of that ... (question)
from Jose to Everyone:
@stian: as a relation?
from Stian and Jits to Everyone:
Khalid: So that a workflow is a version of another, that is up to the creator to make that decission, rather than trying to infer it
from Stian and Jits to Everyone:
@Jose exactly
from Stian and Jits to Everyone:
@Jose - relating the workflow to the RO. (the ontology guys can fight over exactly how to state this)
from Rafa(UPM) to Everyone:
no
from Rafa(UPM) to Everyone:
rafa is Rafa(upm)
from Rafa(UPM) to Everyone:
(smile)
from Pique to Everyone:
user is the final responsible of the info about the content of his RO
from Stian and Jits to Everyone:
Khalid: Should we require that a RO must have one workflow?
from graham Klyne to Everyone:
I don't like the idea of adding arbitrary constraints.
from Jits to Everyone:
@graham - +1
from Stian and Jits to Everyone:
Jose: A RO might not need a WF out side of this project, bu there the wf is a cornerstone of what we want to preserve. So workflows should appear in the RO (as a 1st class citizien)
from Stian and Jits to Everyone:
Marco: Agree with that, within this project that is a requirement
from Stian and Jits to Everyone:
Wf4ever!
from Stian and Jits to Everyone:
Marco: Depends on the domain, in experimental sciences we have at least one methods, but other domains might not even have that
from Stian and Jits to Everyone:
Khalid: In his view a method is an abstract wf
from Stian and Jits to Everyone:
s/his/my/
from graham Klyne to Everyone:
But I would expect there to be ROs without w/f that are used by ROs with w/f, e.g.
from Jose to Everyone:
@stian: mmm, then the wf is regarded as something external (decoupled) from the RO
from Stian and Jits to Everyone:
Khalid: Is a RO just an aggregation, or do we specify a method and a workflow?
from Stian and Jits to Everyone:
Pique: Tend to have same opinion as Marco. Agree with Jose here.
from Stian and Jits to Everyone:
Khalid - something from Jits?
from Stian and Jits to Everyone:
Jits: Follow-on from some emails, Jose asked on Marco's proposal
from Jose to Everyone:
sorry I couldn't attend to the users telco yesterday
from Stian and Jits to Everyone:
we discussed some parallell activity in the user telcon, GK is working on this command line to bridge this, as a kind of mockup tool
from Stian and Jits to Everyone:
Jits building a quick mock interface for doing annotation, just spitting out AO triples or similkar
from Stian and Jits to Everyone:
Jits: Taking Marco's proposal and building something quick, then tie that into the other activities over the next few weeks
from Stian and Jits to Everyone:
Jose: Relating to GK's work on the command line tool, how much in line are these
from Stian and Jits to Everyone:
Jose: we might run the risk of starting parallell lines, difficult to make them converge, but it sounds under control
from Stian and Jits to Everyone:
Jits: Valid concern, in the ARch call we'll talk about this
from Stian and Jits to Everyone:
Jits: Just getting something quick out first for discussion
from graham Klyne to Everyone:
Yes, we've been talking to make sure we don't take divergent paths
from Stian and Jits to Everyone:
Jits: connecting the cmndline tool wioth RObox and the mockup tool
from Stian and Jits to Everyone:
Khalid: Another item
from Stian and Jits to Everyone:
Khalid: Rafa here?
from Stian and Jits to Everyone:
Khalid: Is there naything happening on collaboration/evolution in this aspect?
from Stian and Jits to Everyone:
Raul: First..
from Stian and Jits to Everyone:
Rafa has been on holiday
from Stian and Jits to Everyone:
Raul: Evolution mostly see.. starting looking at lifecycle model. Two questions on this in the deliverable and diagram
from Stian and Jits to Everyone:
Raul: Some feedback from Pique and Graham
from Stian and Jits to Everyone:
overall idea, to build a vocabulary
from Stian and Jits to Everyone:
like evolution ontology
from Stian and Jits to Everyone:
can be compared to evoloution of ontologies
from Stian and Jits to Everyone:
will have a sample model in a few weeks
from Stian and Jits to Everyone:
then we can ellaborate it before the poznan meeting
from Stian and Jits to Everyone:
Khalid: comment on this picture of lifecycle, in my opinion it is not final
from Stian and Jits to Everyone:
issues that you come up with, then we can revisit it
from Stian and Jits to Everyone:
Raul: how the RO can evolve from one state to another, and how changes are represented
from Stian and Jits to Everyone:
between different states and versions
from Stian and Jits to Everyone:
Khalid: On collaboration, an email from Carole, mentioning 1/2 papers by Wei Tain on services
from Stian and Jits to Everyone:
we had a telcon with Wei to hear about his work
from Stian and Jits to Everyone:
they made analysis of community of users, analysis workflows on myExperiment, services
from Stian and Jits to Everyone:
Jose: interesting Discussion on the mailing list about sciientific social object
from Stian and Jits to Everyone:
Khalid: yes; Wei's work is connected to this, what connects to a design of a service will need.. on connection between users between workflows they are using (question)
from Stian and Jits to Everyone:
social dimentions implicitly described
from Stian and Jits to Everyone:
user model in exercise
from Stian and Jits to Everyone:
Jose: analysising scientist interactions and RO/wf interactions, then you could extract a model of what is important in collaboration
from graham Klyne to Everyone:
@stian - just tookj a peek at https://github.com/wf4ever/ro/blob/master/ro-vocab.ttl - I think some of the terms (e.g. ro:hasInput, etc.) should align with the emerging provenance ontology.
from Stian and Jits to Everyone:
@GK totally agree
from Stian and Jits to Everyone:
@GK in arch f2f we said to use OPM for now.. which one?
from Stian and Jits to Everyone:
Jose: In this morning's email from me.. (question)
from Stian and Jits to Everyone:
Jose: making implicit the relations between ROs
from Stian and Jits to Everyone:
@Khalid mute please
from khalid to Everyone:
@Graham, thanks that's a good point, we can try to revise the Provenanec WG model to see the vocabulary that may be useful for us
from khalid to Everyone:
I am muted
from Stian and Jits to Everyone:
thanks (smile)
from Stian and Jits to Everyone:
Marco: Not with workflows, but for publications, we're doing things that are similar to what you are suggesting in text mining methods
from Stian and Jits to Everyone:
experts that do similar things
from Stian and Jits to Everyone:
papers that are similar
from Stian and Jits to Everyone:
mine the methods
from Stian and Jits to Everyone:
methods to be similar
from graham Klyne to Everyone:
@stian - for time being, I'd suggest looking at http://purl.org/net/opmv/ns - it's a simplification of OPM, and it has been developed in conjunction with open gov data developers, so has some work hardening.
from Stian and Jits to Everyone:
@GK, yes, I've looked at that before, OPM seperates the vocabulary from the ontology, right. And Jun should know about it
from Stian and Jits to Everyone:
Khalid: Thanks everyone
from Jose to Everyone:
ack @stian for the minutes!
from Jose to Everyone:
@stain @gk: good point as to the provenance vocabulary

  • No labels