- From: Felix Sasaki <fsasaki@w3.org>
- Date: Mon, 17 Sep 2012 13:04:10 +0200
- To: Olaf-Michael Stefanov <olaf@stefanov.at>
- Cc: Dave Lewis <dave.lewis@cs.tcd.ie>, "public-multilingualweb-lt@w3.org" <public-multilingualweb-lt@w3.org>
- Message-ID: <CAL58czq2MKvOx72gbz1yFZVDSsytTZ03P_pqtGgfTaSVetTBkw@mail.gmail.com>
Thanks for your changes, Olaf I'll leave it to Dave to integrate these as he decides. A PDF probably would be best for others to see your proposals. Best, Felix 2012/9/15 Olaf-Michael Stefanov <olaf@stefanov.at> > Dear Dave and friends, > > Attached is both an MS-Word and OpenOffice version of sections 1-5 of the > text of the link you provided < > http://www.w3.org/International/multilingualweb/lt/wiki/Simple_Segment_Machine_Translation_Use_Case_Demonstration#Summary> > (with track changes in Final-show-mark-up format), with > 1 correction to Section 1, > 5 corrections to Section 2, > 4 corrections to Section 3, and > 3 corrections to Section 5 (whereby I'm a bit unsure if the 1st correction > (replacing "as visible" with "is visible" in the 1st sentence). > > Otherwise I find the text developing very well for its intended purpose. > > Please let me know, for future reference if MS-Word or OpenOffice versions > are preferred. > > Kind regards, > olaf-michael > > > On 2012-09-14 15:14, Dave Lewis wrote: > > Hi Felix, > No problem on summary - done: > > http://www.w3.org/International/multilingualweb/lt/wiki/Simple_Segment_Machine_Translation_Use_Case_Demonstration#Summary > > Dave > > On 14/09/2012 09:06, Felix Sasaki wrote: > > Hi Dave, > > I'm changing the topic since I hope that we can decide on this soon, > since many people already work on use case examples for the "implementation > demo" session 25 September. > > Can we agree on the simple version? You write "A quick summary section > directly giving the benefits as you suggest would definitely be good." - I > just want to be sure that everybody would prepare something like that, > along the lines mentioned below: > > - This is our implementation. > > - The metadata solves the following problems: > > (bullet list). Example for "translate": "translate metadata assures that > pieces of content are not translated > > - Benefits: better translation quality, ... > > Above is basically what you created in the wiki, just scaled down. So > details are fine too, but everybody is busy after the summer break, and > what we currently mostly need are simple example - not for us, but the > people (hopefully) watching us :) > > More comments (not so urgent ;) ) below. > > > 2012/9/14 Dave Lewis <dave.lewis@cs.tcd.ie> > >> Hi Felix, >> Thanks for those suggestions. Currently, I was targetting this at a level >> useful primarily for communication within the WG leading upto Prague. So we >> can then refine these for a more general audience after that - when the >> spec is more stable. >> >> A quick summary section directly giving the benefits as you suggest would >> definitely be good. >> >> By general audience, I guess you still mean someone interested in the >> technical details of interoperability and wanting to understand the >> specific benefits of ITS? So the aim would be to get them reading and >> hopefully implementing (or asking a provider to implement) ITS2.0 - right? >> > > A general audience would IMO be somebody who doesn't know about language > technology, ITS metadata and the tools we are working on - but we want to > convince him that the tools solve a real life problem. > > >> >> We are aiming for a sample online version of CMS LION exactly as you >> suggest - but we don't have a roll-out date yet - a few month off I think. >> Certainly the aim is to have an interesting multiway >> (XLIFF/PROV/ITS/RDF/NIF) interoperability demonstrator for CNGL, rather >> than a product or downloadable library. >> > > > Understand, for TCD as an academic participant in the group / project > that totally makes sense. However, for the industry partners I would hope > that we can get something along the lines of Okapi or ITSTools. The main > point is not open source or not, but re-producability. > > >> >> You are right about the language info not being directly used in the >> scenario, since we use translate to impact the MT behaviour. It was more to >> help us test out the CMS-LION parsing for this. We'll have a bit more of a >> think of a good example for language info - its a bit tricky to think of >> one in HTML5. >> > > The main use case for language information is to map non xml:lang > attributes to the value you would expect xml:lang. That can e.g. support > workflow decisions ("should this content go to MT engine / translator A or > B?"). So if you have an example along those lines in XML, I'm happy to > create an HTML5 version. > > Best, > > Felix > > > >> >> Regards, >> Dave >> >> >> On 12/09/2012 09:01, Felix Sasaki wrote: >> >>> >>> hanks a lot for the template and the example. I would propose to >>> simplify the description a lot. It is too detailed for a general audience. >>> We can add more detailed descriptions in a separate section. But the main >>> section could just consist of short descriptions - max 1 paragraph for each >>> item - saying: >>> >>> - This is our implementation: CMS Lion, Statistical MT System. >>> >>> - The metadata solves the following problems: >>> >>> (bullet list). Example for "translate": "translate metadata assures that >>> pieces of content are not translated >>> >>> - Benefits: better translation quality, ... >>> >>> - Example. >>> >>> >> >> > > > -- > Felix Sasaki > DFKI / W3C Fellow > > > > -- Felix Sasaki DFKI / W3C Fellow
Received on Monday, 17 September 2012 11:04:42 UTC