[ome-devel] OME-XML Internationalization

[Josiah Johnston]

I've started working on OME internationalization with Javier Tibau  
and Daniel Ochoa from the Center of Vision and Robotics in Guayaquil,  
Ecuador. They wish to provide a functional web interface in spanish  
for APROFE (translation: The Association to Promote the Health of  
Ecuadorian Families). APROFE is a non-profit health care provider,  
wishes to use OME to expedite communication between their technicians  
and pathologists.

I'm writing this email to give folks a heads-up, and to probe the  
level of interest. A high level of support for internationalization  
in the OME-XML interchange standards will be superfluous unless there  
are people working in different languages who wish to exchange data.  
Our current goal is to get a working prototype with minimal branching  
from OME-HEAD. There's four phases to this:
1) Semantic Type labels and descriptions in multiple languages  
(support in OME-XML and the database)
2) Finishing making the web UI completely template-driven, and  
provide translations of templates
3) Specifying the language an attribute (ST instance) is written in  
(support in OME-XML and the database)
4) Supporting non-english character sets (i.e. UNICODE) in OME-XML  
and the database

We're currently focusing solely on 1 and 2. For this immediate  
project, 3 and 4 are unnecessary because they wish to use OME for  
domestic communication, and an english character set sufficiently  
approximates the spanish character set for a functional system.

The proposed changes to Semantic Type Definitions are minor, and  
could be easily ignored by legacy agents. The changes are to allow  
many "Label" tags under an ST or Semantic Element (SE), each of which  
would specify an "xml:lang" attribute. For legacy purposes, the  
"Description" immediately under SemanticType would be assumed to be  
english. This proposal hasn't been included in the OME-XML evolution  
page because the implementation and level of planning is still  
slightly too raw. I think UNICODE support (4) would be required  
before it can be part of a standard, and we're several months away  
from investigating that. Example:

<SemanticType Name= "Pixels" AppliesTo = "I">
	<Description>Storage location and data type of the image pixels,  
including the extent of each dimension in the 5-D array.</Description>
	<Label xml:lang="es" Name="Pixeles">
		<Description>Localizacion del almacenaje y tipo de datos de los  
pixeles de la imagen, incluyendo el grado de cada dimension en el  
arsenal 5-D.</Description>
	</Label>
	...
	<SemanticElement ...>...
</SemanticType>

Unfortunately, there can be redundancy in description if there is an  
english label. The label itself will not be redundant, because it  
will allow a human formatted version of an ST. For example,  
DisplayOptions' label could include a space.
I considered having the translated Descriptions under ST instead of  
Label to avoid potential for redundancy. However, that could impact  
existing implementations that assume there are exactly 0 or 1  
"Description" children of SemanticType.

Is anyone else considering internationalization? I've seen a bit of  
traffic on the list from non-english speaking countries. Is English,  
the new Latin, sufficient for your needs?
Has anyone outside of the core OME team implemented support for  
semantic type definitions? If so, do you foresee these changes  
screwing up your code?

-Josiah