[Standards] UPDATED: XEP-0171 (Language Translation)

[Fletcher, Boyd C. CIV US USJFCOM JFL J9935]

There are ways to detect which lang is being used based on the encoding and word choice and some statistical measurements of the text so its possible to detect automaticallly. Its not 100% but is pretty accurate for mainstream proper langs. Its gets very complex when you deal with dialects. For example, in arabic there is the formal modern standard arabic (.msa) but its not the main lang of most arabic speakers - they speak dialects but some have significant differences from msa.. Detecting them automatically can be tricky. Similar problems exist in all langs.



-----Original Message-----
From: standards-bounces at xmpp.org <standards-bounces at xmpp.org>
To: XMPP Extension Discussion List <standards at xmpp.org>
Sent: Mon Feb 18 09:10:57 2008
Subject: Re: [Standards] UPDATED: XEP-0171 (Language Translation)

Well, if sending user does not want to set message language (which is 
understandable, as it is awkward solution), then maybe receiving user 
will do that?
You can provide a way for user to set up translation for any of his 
contacts. I.e. define one (or more) languages that could be used by this 
contact.
Then translator can translate messages from all these languages to 
user-defined language.
User can define one or more languages he understands (with priorities; 
translator would also have it's own priority table to try to translate 
between more similar languages if possible; i.e. if i understand English 
and Russian, then translator will translate from German to English, 
because it's easier).
As i said, translator performs translation for all source languages and 
sends all translated text to user (this is arguable, though, since CPU 
time consumption and bandwidth consumption will increase significantly 
if there is more than one source language). User then can switch 
translations just as he would switch character encoding tables in 
browser: if text is unreadable, then he should try another translation. 
User's client may also use elements of AI to guess which translation is 
the best.
Sender may also specify which language he uses in his messages, but this 
is only a hint for translator.