[Obi-devel] proposed checks and corrections on branch files

[Melanie Courtot]

Hi all,

For the next steps of the release process we would like to add some features allowing for checking and corrections of our branch files, and especially our metadata.

Here is the list of proposed modifications: if you have any objection to those let us know, as otherwise we will propose that they be approved at the next dev call (2 weeks - April 23rd)

1. curation status annotation property

A curation status annotation property must be present on each term, and if it's value is empty it will be defaulted to "uncurated" by the script.

2. underscore in label
Our labels and preferred term should be space separated. Underscores in labels and preferred terms will be removed by the script (except for annotation properties themselves)

3. rdfs:label and obi:preferred_term
These will be synchronized, i.e. we will check that each term has both fields and that they are identical

4. rdfs:comment and obi:definition
We propose to synchronize these as well: each term should have both and they should be identical. At the moment OBI doesn't make use of the rdfs:comment field, but we will check for pre-existing rdfs:comment before synchronization though to not loose annotations that would have been entered in the rdfs:comment.

Synchronizing rdfs:comment/obi:definition and rdfs:label/obi:preferred_term are for the benefit of tools which understand the roles of the former and not the latter. In a future version of OWL it may be the case that we can simply make obi:definition a sub property of rdfs:comment and avoid the explicit duplication. We don't expect this duplication to substantially effect the size of the file or performance of tools using the ontology.

We know that there are other checks and modifications that could be done, this is the first step.

Thanks,
The OBI release team

[sch...@ebi.ac.uk]

Hello all,
see inline...

> Hi all,
>
> For the next steps of the release process we would like to add some
> features
> allowing for checking and corrections of our branch files, and especially
> our metadata.
>
> Here is the list of proposed modifications: if you have any objection to
> those let us know, as otherwise we will propose that they be approved at
> the
> next dev call (2 weeks - April 23rd)
>

> 1. curation status annotation propertyA curation status annotation

> property
> must be present on each term, and if it's value is empty it will be
> defaulted to "uncurated" by the script.
>
> 2. underscore in label
> Our labels and preferred term should be space separated. Underscores in
> labels and preferred terms will be removed by the script (except for
> annotation properties themselves)

we better mark the underscore-indicated helperclasses before substituting
it with space filewide. Or just don#t substitute where underscore is a the
beginning of a name.

>
> 3. rdfs:label and obi:preferred_term
> These will be synchronized, i.e. we will check that each term has both
> fields and that they are identical
>
> 4. rdfs:comment and obi:definition
> We propose to synchronize these as well: each term should have both and
> they
> should be identical. At the moment OBI doesn't make use of the
> rdfs:comment
> field, but we will check for pre-existing rdfs:comment before
> synchronization though to not loose annotations that would have been
> entered
> in the rdfs:comment.

For these (3. and 4.) we should probably decide on one, which we fill out
as editors, and then later autogenerate the other ... editors should not
be required to put in the redundant names and definitions in the
corresponding fields. So either
a)just use our annotation properties and make these the defaut display
key. Populate the homologous rdfs fields at release time.
or
b)use rdfs:label and rdfs:comment (label as display key) and populate our
annotation properties at release time.
I tend to prefer alternative a.
Cheers,
Daniel Schober

>
> Synchronizing rdfs:comment/obi:definition and
> rdfs:label/obi:preferred_term
> are for the benefit of tools which understand the roles of the former and
> not the latter. In a future version of OWL it may be the case that we can
> simply make obi:definition a sub property of rdfs:comment and avoid the
> explicit duplication. We don't expect this duplication to substantially
> effect the size of the file or performance of tools using the ontology.
>
> We know that there are other checks and modifications that could be done,
> this is the first step.
>
> Thanks,
> The OBI release team

> -------------------------------------------------------------------------
> This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
> Don't miss this year's exciting event. There's still time to save $100.
> Use priority code J8TL2D2.
> http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone_______________________________________________
> Obi-devel mailing list
> Obi-...@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/obi-devel
>

-------------------------------------------------------------------------
This SF.net email is sponsored by the 2008 JavaOne(SM) Conference
Don't miss this year's exciting event. There's still time to save $100.
Use priority code J8TL2D2.
http://ad.doubleclick.net/clk;198757673;13503038;p?http://java.sun.com/javaone
_______________________________________________
Obi-devel mailing list
Obi-...@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/obi-devel