[MOBY-dev] xml:lang explored

Wed May 28 10:30:09 UTC 2008

Hey All,

I have heard three ideas:

1) allow registration with multiple languages per instance
2) stick with english only
3) register with one language, but record the language

Of these cases I believe that using xml:lang is only valuable if 1) is chosen.

My understanding of the i18n support in general and xml:lang in
particular is to support a user setting his/her preferred language and
getting data in that language if it is available. So if the Dashboard
allowed us to set our preferred language to German and then retrieved
descriptions in German or fell back to English then I think we could
argue the effort was valuable.

But for case 3) I feel it would be a waste of effort - it wouldn't
change the existing user experience - it would just put a label on
what we already know. I don't believe users would search for services
with Spanish-only descriptions - but I do believe they would ask for
descriptions in Spanish if they existed.

I'm not here to advise whether we should or shouldn't support xml:lang
- only to advise that my personal opinion is that level of re-coding
is only useful if we support multiple descriptions in different
languages.

Personally I don't think there is a high demand in bioinformatics for
this feature - so I would advise against it. On the other hand - Mark
has pointed out that he hopes this work will have impact outside
bioinformataics. I can also add that the amount of code change is
pretty small for MOBY Central - the real impact is the change in the
clients to set language preferences.

So I am torn. I'm happy to make the change in MOBY Central - but it
won't be useful until the clients support it.

What do people say?

Cheers, jas.

2008/5/27 Paul Gordon <gordonp at ucalgary.ca>:
> I will reiterate that I think the language of the text should be noted in
> the database somewhere, but that we shouldn't support multi-liguality for a
> single entry (that's too much work).  While we can pressure people into
> using English for the public central registry, I see no reason to force
> English on a boutique registry in Botswana (for example) just because we
> were too lazy to program in an extra column in the DB.  I am also thinking
> about the potential usefulness of Moby outside of science...
>
> Mark Wilkinson wrote:
>>
>> personally, I tend to think we *should* be using the multi-lingual
>> capabilities of XML more than we are!  The idea is not to introduce a tower
>> of Babel, but rather to embrace as many languages as possible in our
>> infrastructure!  to be inclusive, rather than exclusive!
>>
>> this does, however, require extensive re-tooling of the software.  I
>> wonder if it is something that is best left for Moby 2.0 (and planned from
>> the start) rather than hacked into Moby 1.0?
>>
>> M
>>
>>
>>
>>
>> On Mon, 26 May 2008 15:41:34 -0700, Pieter Neerincx
>> <pieter.neerincx at gmail.com> wrote:
>>
>>> Hi all,
>>>
>>> On 26 May 2008, at 20:59, Paul Gordon wrote:
>>>
>>>> I think we are mostly polyglots on this list,
>>>
>>> I'm a polyglot when I'm on holiday or in a pub maybe, but when I'm
>>> talking bioinformatics stuff I'm a strict monoglot! Before we open that can
>>> of worms: do we actually need translations? Did anyone request support for
>>> different languages? Mark mentioned someone did on the last developer
>>> meeting, but I didn't see anyone chime in on this list with a request for
>>> other languages. I only noticed a few objects and services in the official
>>> public Central containing descriptions in Spanish (or is it Portugese?),
>>> which renders them useless for the majority of users.
>>>
>>> BioMoby web services are all about creating interoperable resources.
>>> Introducing the Tower of Babel effect isn't going to help. I hope I don't
>>> sound arrogant, but unless there are scientists out there using BioMoby for
>>> research into language itself, archeology or culture, I really fail to see
>>> why we need anything but the "de facto" standard language for research. (No,
>>> I'm not a native Englisch speaker...)
>>>
>>> Is there actually any major bioinformatics resource that supports
>>> translations? I can't think of any, but maybe I should write an e-mail to
>>> the NCBI, DDBJ, EBI et al. and request Dutch translations of Genbank,
>>> PubMed, KEGG, Uniprot, Ensembl, etc. Might be fun to see if and how they
>>> repsond :)...
>>>
>>>> but for the sake of technical simplicity, I'd stick with just tracking
>>>> the language rather than having multiple values in the registry.
>>>
>>> For even more simplicity I'd opt for hardcoding the language to English.
>>> Now let's hope the native English speakers are not going to fight over
>>> whether international English means Australian, UK, American, Canadian, or
>>> yet another English. If that happens I'm voting for Dinglish :).
>>>
>>> Just my € 0.02
>>>
>>> Cheers,
>>>
>>> Pi
>>>
>>>>  Multilingual descriptions could be offloaded to the LSID metadata
>>>> perhaps if people really want it.
>>>>
>>>> Mes 0.02$ canadiens,
>>>>
>>>> Paul
>>>>
>>>> Mark Wilkinson wrote:
>>>>>
>>>>> Should we allow registration of objects, service-types and namespaces
>>>>> in foreign languages also?  If so, then we need to re-think the entire way
>>>>> we manage the ontology, and assign unique id numbers to each node, where the
>>>>> rdf:label of the node can have multiple languages, rather than having the
>>>>> node named by its label.
>>>>>
>>>>> ...can... worms... but it's probably the "right thing to do"...
>>>>>
>>>>> M
>>>>>
>>>>>
>>>>>
>>>>> On Mon, 26 May 2008 00:21:23 -0700, Jason Stewart
>>>>> <jason.e.stewart at gmail.com> wrote:
>>>>>
>>>>>> Hey all,
>>>>>>
>>>>>> Here's what I have found out.
>>>>>>
>>>>>> I don't believe that changing the current parser for Central.pm is
>>>>>> going to help the situation. I have looked and besides special API's
>>>>>> like the one implemented by LibXML::Reader, the application is always
>>>>>> required to maintain the status of xml:lang using the standard SAX and
>>>>>> DOM API's - for Java or for Perl.
>>>>>>
>>>>>> We will have to preserve the information in the DB, so a decision
>>>>>> needs to be made on what level the xml:lang should be used - only for
>>>>>> descriptions?? or for the whole registration?? At the moment I don't
>>>>>> see any reason to use it for more than the description - how do people
>>>>>> feel about this? Changing this should require adding one column to the
>>>>>> DB.
>>>>>>
>>>>>> Finally, do we want registration with multiple descriptions? That will
>>>>>> probably require a significant change in the DB - a new table for
>>>>>> linking descriptions to registrations. Are people happy about that?
>>>>>>
>>>>>> Cheers, jas.
>>>>>> _______________________________________________
>>>>>> MOBY-dev mailing list
>>>>>> MOBY-dev at lists.open-bio.org
>>>>>> http://lists.open-bio.org/mailman/listinfo/moby-dev
>>>>>
>>>>>
>>>>>
>>>> _______________________________________________
>>>> MOBY-dev mailing list
>>>> MOBY-dev at lists.open-bio.org
>>>> http://lists.open-bio.org/mailman/listinfo/moby-dev
>>>
>>>
>>> _______________________________________________
>>> MOBY-dev mailing list
>>> MOBY-dev at lists.open-bio.org
>>> http://lists.open-bio.org/mailman/listinfo/moby-dev
>>
>>
>>
> _______________________________________________
> MOBY-dev mailing list
> MOBY-dev at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/moby-dev
>