- From: Jacob Palme <jpalme@dsv.su.se>
- Date: Mon, 9 Sep 2002 10:02:31 +0200
- To: discuss@apps.ietf.org
- Message-Id: <p0510030db9a1fc5ed90c@[192.168.100.102]>
What is the best architecture for machine and/or human
language translation in e-mail?
Here are some possible architectures:
(1a) Sender translates message to multiple languages,
and sends the message as multipart/alternative or
some variation of this, with a separate body part
for each language.
(1b) Sender knows the language preferences of each
recipient, and send in the language(s) preferred
by each recipient. Note that some recipients may
want multiple languages, for example both
original in the original language and a translation
to the native language of the recipient.
(2) Sender sends the message to a translation agent,
which forwards the message to recipients as
1a or 1b above.
(3) Mailing list expander translates and forwards
as 1a or 1b above.
(4) Recipient uses a human or machine translator
to translate on receipt.
Are there more architectures which I have forgotten
above?
Some pros and cons of the architectures:
+-------------+--------+--------+--------+--------+--------+--------+--------+
|Aspect |1a |1b |2a |2b |3a |3b |4 |
+-------------+--------+--------+--------+--------+--------+--------+--------+
|Translation |Sender |Sender |MT agent|MT agent|List |List |Reci- |
|at | | | | |expander|expander|pient |
+-------------+--------+--------+--------+--------+--------+--------+--------+
|Registry or |No |Yes |No |Yes |No |Yes |No |
|negotiation | | | | | | | |
|of user prefs| | | | | | | |
+-------------+--------+--------+--------+--------+--------+--------+--------+
|Sending |Yes |No |Yes |No |Yes |No |No |
|multiple | | | | | | | |
|copies of | | | | | | | |
|message | | | | | | | |
+-------------+--------+--------+--------+--------+--------+--------+--------+
|Cost of |Lower |Lower |Lowest |Lowest |Lowest |Lowest |Higher |
|translation | | | | | | | |
+-------------+--------+--------+--------+--------+--------+--------+--------+
|Manual |Yes, if |Yes, if |Yes, if |Yes, if |Yes, if |Yes, if |Expen- |
|translation |"Super- |"Super- |"Super- |"Super- |"Super- |"Super- |sive: |
|can replace |sedes" |sedes" |sedes" |sedes" |sedes" |sedes" |Duplicat|
|machine |e-mail |e-mail |e-mail |e-mail |e-mail |e-mail |e work |
|translation |header |header |header |header |header |header | |
|if needed |gets |gets |gets |gets |gets |gets | |
| |accepted|accepted|accepted|accepted|accepted|accepted| |
+-------------+--------+--------+--------+--------+--------+--------+--------+
|Quotes of |Requires|Impossi-|Requires|Impossi-|Requires|Impossi-|Only |
|previous |complex |ble |complex |ble |complex |ble |partly |
|messages in |protocol| |protocol| |protocol| |possible|
|reply should | | | | | | | |
|not get | | | | | | | |
|translated | | | | | | | |
|back and | | | | | | | |
|forward | | | | | | | |
+-------------+--------+--------+--------+--------+--------+--------+--------+
|Compatibility|Sender |Sender |Yes |Yes |Yes |Yes |Reci- |
|with existing|needs |needs | | | | |pient |
|mail programs|plugin |plugin | | | | |needs |
| | | | | | | |plugin |
+-------------+--------+--------+--------+--------+--------+--------+--------+
|New protocol |Multi- |Multi- |Multi- |Multi- |Multi- |Multi- |None |
|needed |part |part |part |part |part |part | |
| |usage |usage |usage, |usage |usage |usage | |
| | |and |trans- |and | |and | |
| | |content |lation |content | |content | |
| | |negoti- |request |negoti-a| |negoti- | |
| | |ations | |tions, | |ations | |
| | | | |trans- | | | |
| | | | |lation | | | |
| | | | |request | | | |
+-------------+--------+--------+--------+--------+--------+--------+--------+
--
Jacob Palme Stockholm University
Professor and KTH Technical University
E-mail: jpalme@dsv.su.se WWW: http://www.dsv.su.se/jpalme/
Snail mail: Skeppargatan 73 Personal phone: +46-8-16 16 67
S-115 30 Stockholm, Sweden Personal fax: +46-8-783 08 29
Department of Computer and Systems Sciences,
Research group on Computer Mediated Communication,
WWW: http://www.dsv.su.se/jpalme/cmc-research-at-DSV.html
Received on Monday, 9 September 2002 04:18:38 UTC