Re: Minutes [Re: ML Schema call agenda May 8, 2017]

Hello Joaquin,

On Wed, May 10, 2017 at 11:56 PM, Joaquin Vanschoren <j.vanschoren@tue.nl>
wrote:

> Great news, but I didn't understand everything :). Which 'lib' do you mean
> exactly? SML Bench? Does it do more than export its benchmarks to MLSchema?
>

- ​SML Bench is just a use case, regardless schema (at this moment is MEX
based)
- This library would make sub-schemata (i.e., MEX, DMOP, ....) be
interoperable based on ML-Schema mappings.
- Thus, any technology/software that implement ANY of the schemata, would
automatically work with ML-Schema and ALL others sub-schemata (through this
library)

In the end, if someone releases a software "A" that just works with *Expose*,
we would have also *MEX* (as output metadata) "for free" and vice-versa. In
case we have "B" working with MEX (and A and B have the same objective),
the output metadata can be then, after conversion, be comparable and
visualized "side-by-side". It's just like a script for data conversion.
​

>
> Also, my impression was that MLSchema is a high-level schema and that MEX,
> OntoDM, DMOP,... are more specific extensions.
>

​Yes,

the canonical model of the architecture.


> How would you 'export' between them?
>

​By using this library, basically it will have the mappings of the schemata
and all the rules for data conversion. Also, if you have, let's say, a REST
service to import/export to OpenML and we integrate such feature (lib) into
this REST, we could also benefit of more data outputs.

Example:

- library.convert("mex-example.ttl", "ontodm.rdf", "MEX", "OntoDM")
would read mex-example.ttl and convert to OntoDM schema (output ontodm.rdf)
based on MLSchema mappings.
we could also potentially have:
- library.convert("mlschema.rdf", "mex-example.ttl", "MLSchema", "MEX")
that would read mlschema file and export to MEX file

Of course we have to check some constraints, but since we have a overlap
(supported by Agnieszka's work [1]) among the sub-schemata, that is
achievable.

​

> Even for OpenML, the RDF we export goes beyond what is defined (purely
> speaking) in MLSchema (@Tomasso, correct me if I'm wrong?).
>

​That is what I assume :)
​

> I guess the best you can do is make MLSchema rich enough so that you can
> at least export the main common information? Or maybe I'm not grasping your
> point :).
>

​I think we will be sure about that (making OpenML the central repository
for all others) as soon as we finish the mappings (actually we started this
conceptual alignment before). So we would have two main options: either 1)
define a new "feature" in OpenML to import the basic (common) information
or 2) we extend ML-Schema to cover 100% of OpenML.
At this moment I would vote for (1) since

- It is more simple and straightforward
- we keep MLSchema as simple as possible, otherwise we have a higher risk
of developing something too complex that is likely to not be useful for
everybody

Am I clear in my thoughts? :)
​
[1] http://ontologydesignpatterns.org/wiki/images/4/41/WOP2016_paper_07.pdf


> Cheers,
> Joaquin
>
>
> On Wed, May 10, 2017 at 11:40 PM Diego Esteves <diegoesteves@gmail.com>
> wrote:
>
>> Hi all,
>>
>> let me add some more observations, please:
>>
>> SML Bench [1] has been released with MEX support, which means +1 use case
>> for us!
>>
>> Thus, as soon as we develop the lib to achieve the "horizontal
>> interoperability" level (i.e., among ML/DM schemata), automatically
>> benefiting from that. As a consequence, more and more data can be (in a
>> transparent manner) feed into OpenML.
>>
>> This kind of 2-dimensions levels are like the 5 starts of Semantic Web =)
>> We need it!
>>
>> Clear advantages are:
>>
>> - support our future plans
>> - support project proposals
>> - and further increase the visibility of the group
>> - it is very compelling to the end-user. As we all know, a major issue
>> with workflow systems, for instance, you can only interchange data between
>> users of the same system.
>>
>> @Joaquin,
>> I was wondering that maybe you have some sort of interface for OpenML
>> anyway and we could merge the project design to consider the development of
>> this scenario (this lib to export [MEX<->OntoDM<->Expose<->DMOP]<=>
>> MLSchema). There was a guy working in a development of a REST interface
>> for MEX also, I can check that.
>>
>> @Ciro,
>> what about WASOTA plans? Why don't we simply integrate to OpenML and
>> start interchanging data using this REST interface, which could implement
>> this library we are going to design and release?
>>
>> This library project means hitting MANY MANY MANY targets with one shot!
>> ;)
>>
>> @Agnieska, all
>> Finally, a journal submission would also boost visibility. I think all
>> the concepts are mature enough.
>>
>> Cheers,
>> Esteves
>>
>> [1] https://github.com/AKSW/SML-Bench/releases/tag/0.2
>>
>> *Diego Esteves *
>> SDA (Smart Data Analytics) Research Group
>>
>> *“Inspiration exists, but it has to find you working.” Pablo Picasso*
>>
>> On Wed, May 10, 2017 at 1:46 PM, Agnieszka Ławrynowicz <
>> agnieszka.lawrynowicz@cs.put.poznan.pl> wrote:
>>
>>> Dear All,
>>>
>>> Thank you for the call.
>>> Below you can find the minutes from the call also included at [1]:
>>>
>>> Actions:
>>> * OpenML use case:
>>> a) @Joaquin will meet Tuesday 16 with @Tommaso to push forward calling a
>>> script of Tommaso (is not integrated yet to the OpenML platform)
>>> b) @Agnieszka will check whether there is a mapping of ML Schema to Open
>>> ML (should be used both)
>>>
>>> * All: put related papers to the reproducible research Mendeley group
>>> created by @Diego
>>> * @Agnieszka: will talk about use cases in scientific publishing during
>>> ESWC 2017
>>> * @Agnieszka: to talk to editors of the Semantic Web Journal
>>> * @Agnieszka: will check possible uses cases related to Research Objects
>>> * @Larisa: to contact Data Science journal editors
>>>
>>> Suggestions:
>>> * @Larisa: maybe form an Advisory Board and ask them for endorsement
>>> * @Diego suggests to setup a deadline for producing a tool for
>>> transforming between schemas (till August?), we need tools
>>>
>>> The next call is May, 22 2017, 1.30pm CET.
>>>
>>> Best regards and cheers,
>>> Agnieszka
>>>
>>> [1] https://github.com/ML-Schema/core/wiki/Call-minutes
>>>
>>>
>>> Wiadomość napisana przez Agnieszka Ławrynowicz <
>>> agnieszka.lawrynowicz@cs.put.poznan.pl> w dniu 08.05.2017, o godz.
>>> 13:25:
>>>
>>> Dear All,
>>>
>>> Below please find an agenda for today’s call of the Machine Learning
>>> Schema W3C Community Group.
>>>
>>> *Call May 8, 2017, 1.30pm CET*
>>>
>>> Agenda
>>>
>>> 1. Extending ML Schema beyond the core schema.
>>> 2. ML-Schema & Open ML.
>>>
>>> We will use Google Hangout for the call [1].
>>>
>>> Regards and Cheers,
>>>
>>> Agnieszka
>>>
>>>
>>> [1] https://hangouts.google.com/call/dskyydrmpsybpvxhql5aehiht4a
>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups "MLW3C" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to mlw3c+unsubscribe@googlegroups.com.
>>> To post to this group, send email to mlw3c@googlegroups.com.
>>> To view this discussion on the web visit https://groups.google.com/d/
>>> msgid/mlw3c/A77873DF-555E-446E-9800-DB1E0CC02A5A%40cs.put.poznan.pl
>>> <https://groups.google.com/d/msgid/mlw3c/A77873DF-555E-446E-9800-DB1E0CC02A5A%40cs.put.poznan.pl?utm_medium=email&utm_source=footer>
>>> .
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups "MLW3C" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to mlw3c+unsubscribe@googlegroups.com.
>>> To post to this group, send email to mlw3c@googlegroups.com.
>>> To view this discussion on the web visit https://groups.google.com/d/
>>> msgid/mlw3c/2F0A44D2-0AA0-4B03-8D2F-B10EBE5FDC3F%40cs.put.poznan.pl
>>> <https://groups.google.com/d/msgid/mlw3c/2F0A44D2-0AA0-4B03-8D2F-B10EBE5FDC3F%40cs.put.poznan.pl?utm_medium=email&utm_source=footer>
>>> .
>>> For more options, visit https://groups.google.com/d/optout.
>>>
>> --
> Thank you,
> Joaquin
>

Received on Thursday, 11 May 2017 10:37:51 UTC