- From: Felix Sasaki <fsasaki@w3.org>
- Date: Thu, 13 Jun 2013 04:28:15 +0200
- To: Arle Lommel <arle.lommel@dfki.de>
- CC: joerg@bioloom.de, Jirka Kosek <jirka@kosek.cz>, public-i18n-its-ig@w3.org, kim_harris@textform.com, Hans Uszkoreit <uszkoreit@dfki.de>, Aljoscha Burchardt <aljoscha.burchardt@dfki.de>
Hi Arle, all, Am 11.06.13 22:19, schrieb Arle Lommel: > Hi all, > > Regarding Felix's issues #1 (expressivity) and #3 (mapping), see below: > > #1 (EXPRESSIVITY) > The primary loss in going from MQM to ITS is simply that there is much more (potential) granularity in MQM than in ITS and that there are a number of issue types in MQM with no correlate in ITS. For example, the "Verity" category in MQM deals with how well information in a text corresponds with the world of the text. For example, if a localized text mentions that a ground wire will be bare copper but in the target locale is is actually a blue insulated wire, this is a verity error, even if the translation says "bare copper" (in the target language) and is otherwise accurate. There are a number of things like this that map to "other" in ITS 2.0. > > However, it should be noted that nobody is expected to use the full MQM issue-type list and most MQM implementations would use a MUCH smaller subset, so for any given MQM implementation, the mapping is likely to be pretty simple/straight-forward. That's good to know. However, I think we need to go now the next step and try to describe the mapping in an implementable manner. Your #3 mapping helps a lot with that, but again, having the MWM definition of the types avail. would be very helpful. Also, data using MQM that has already been produced would be nice to see - it helps to work by example, also for testing the mapping. You gave me personally a draft of the MQM description a while ago, so having this list looking at this might be helpful. > > > > #3 (MAPPING) > I've updated the mapping table: > > http://www.w3.org/International/its/wiki/Tool_specific_mappings#Mappings_for_Localization_Quality_Issue_Types > > It's actually not too bad now and could easily be automated (going from MQM to ITS) and I have made a recommended default reverse (ITS to MQM mapping). Some of the MQM complexity is hidden in the mapping because I hid some of the granularity under "[+]" symbols. Very cool! Best, Felix > > -Arle
Received on Thursday, 13 June 2013 02:28:50 UTC