History: CLWE 2008 article TODOs

Source of version: 10

Use this page to keep track of things to do.

!Submission deadline: May 3rd!!

---

!Start keeping track of how many pages we have. 

Limit is 15 pages. Is there an easy way to render these wiki pages using the WS08 Latex format, so we can see how many pages it gives?

Formats are on this page:

   http://www.acm.org/sigs/publications/proceedings-templates

The WS08 CFP does not say if we have can use Option 2 (Tighter Alternate style which may reduce the total number of pages). I (AD) say we go for Strict adherence to ACM SIG, and if we find we are over 15 pages, then use the tighter version. If the WS08 complains, we can later reduce its length to fit the Strict adherance template.

---

!Need a consistent good vocabulary to describe things

For example, "linguistic variant" to refer to something like "The English version of page X". This is better than using "version" which is ambiguous (it could refer to something like v1.2 of the page versus v1.3 of the page). Also, I used the term "edit" instead of "content element". I think it more correctly describes what it is. 

We need to make sure we use the vocabulary consistently throughout the paper.

---

!Spell check the whole thing

At the very end.

---

!Talk about scenarios of "cross-translation"

Make sure that the Storyboard includes a case like this.

Here's a description of how this can happen:

~np~
[12:05:43 PM] Alain Désilets says: EN = {e1}

[12:05:47 PM] Alain Désilets says: FR = {e2}

[12:06:07 PM] Alain Désilets says: FR importe e1 de EN.

[12:06:09 PM] Alain Désilets says: on a

[12:06:14 PM] Alain Désilets says: EN = {e1}

[12:06:18 PM] Alain Désilets says: FR = {e1, e2}

[12:08:12 PM] Alain Désilets says: EN veut importer e2 de FR.

[12:08:20 PM] Alain Désilets says: mais quand il le fait, il va voir et e1 et e2

[12:09:21 PM] Alain Désilets says: Le système sait que EN a déjà e1. Mais le diff n'est pas capable de l'isoler.

[12:10:01 PM] Alain Désilets says: On pourrait pas faire un diff de fr entre la révision ou e2 est apparue en premier, et la révision précédente.

[12:10:44 PM] Alain Désilets says: Mais le problème, c'est comment présenter ce diff de e2 dans un "inline diff" pour la version de FR qui contient maintenant et e1 et e2.

[12:10:53 PM] Louis-Philippe Huberdeau says: Oui. Ici, on a un cas simple qui le montre, mais il y a des cas beaucoup plus complexes
~/np~

---
---

!Misc content removed from the Storyboard section 

Some of it may be re-threaded in the present argumentation.

!!Old Scenario by LPH

!!!The screenshots of this scenario are using the most recent version of CLWE. But there are some key screens missing if we want to tell the full story

In order to illustrate how CLWE supports unconstrained collaborative authoring and editing, we now describe a detailed usage storyboard. This storyboard involves three people (JohnDoe, MarieQuidam and CarloDelPueblo) collaboratively writing a technical report for a project in 3 languages at once (English, French and Spanish). For simplicity sake, we assume all three of them to be trilingual but that each of them are only changing the page in their native language. But note that the system does not require that to be the case.

First, John Doe creates an English page in two consecutive edits. 

{img src=http://wiki-translation.com/tiki-download_wiki_attachment.php?attId=10 align= desc=en_v1}
{img src=http://wiki-translation.com/tiki-download_wiki_attachment.php?attId=11 align= desc=en_v2}

At this point, we have:

  en_v2 = {e1, e2}
  fr = None
  es = None

At this point, Marie Quidam and Juan del Pueblo translate the English page to French and Spanish respectively. __NOTE from AD: It's too bad that we don't have screen shot of what the translation UI looks like, the Under translation notices etc...__ 

{img src=http://wiki-translation.com/tiki-download_wiki_attachment.php?attId=17 align= desc=fr_v1}
{img src=http://wiki-translation.com/tiki-download_wiki_attachment.php?attId=14 align= desc=fr_v2}

At this point, we have:

  en_v2 = {e1, e2}
  fr_v1 = {e1, e2}
  es_v1 = {e1, e2}

And all pages are displayed as being up to date. At this point, Marie adds a list of bullet points to the French version. __NOTE AD: Too bad we don't have a screen shot of Frenh with the bulet points__. At this point, we have:

  en_v2 = {e1, e2}
  fr_v2 = {e1, e2, e3}
  es_v1 = {e1, e2}

Both English and Spanish are displayed as being out of date (''"22% up to date"'') like thus:

{img src=http://wiki-translation.com/tiki-download_wiki_attachment.php?attId=21 align= desc=en_v3_2}

and they provide links to view the "better" French version, and update from it (the green double arrow icon). 

At this point, John decides to update the English version from the French, by clicking on the double green arrow icon in front of the French page name.

{img src=http://wiki-translation.com/tiki-download_wiki_attachment.php?attId=20 align= desc=fr_v2_to_en_v4}

The system shows John the French changes that need to be reproduced in English (changes highlighted in green). John copies and pastes them to the Englsh edit field (__not visible here... which is unfortunate__), and starts overwriting them with their English translation. He then saves it.

At this point, we have:

  en_v3 = {e1, e2, e3}
  fr_v2 = {e1, e2, e3}
  es_v1 = {e1, e2}
 
The English is now shown as being up to date, but the Spanish is still behind. But Juan who looks after the Spanish version is not in the mood for translating right now, and wants to add original content instead. No problem, he can do that. He adds the exact dates of the event along with the location {es_v2}. The page indicates that the page does contain additional content. However, more content can be obtained from the French and English versions. __NOTE from AD: Hum... seems that the image for this transaction is wrong. I don't see dates added!__

We now have:

  en_v3 = {e1, e2, e3}
  fr_v2 = {e1, e2, e3}
  es_v1 = {e1, e2, e4}

and all three versions are considered to need updating. Now that Juan has gotten his original edits out of the way, he is in the mood for translating changes from English. After he is done, we have:

  en_v3 = {e1, e2, e3}
  fr_v2 = {e1, e2, e3}
  es_v1 = {e1, e2, e3, e4}

which means that Spanish is completely up to date, but French and English need updating from Spanish.



__Images below need to be threaded in the above discussion__

{img src=http://wiki-translation.com/tiki-download_wiki_attachment.php?attId=13 align= desc=en_v4}
{img src=http://wiki-translation.com/tiki-download_wiki_attachment.php?attId=15 align= desc=es_v2}
{img src=http://wiki-translation.com/tiki-download_wiki_attachment.php?attId=16 align= desc=es_v3}

As it can be seen in the scenario, to a content contributor, the translation process is invisible. As any visitor of the website, the contributor will see the "Page Translation" box presenting the different alternatives and status information. However, he is free to ignore it. When a change is made by a content contributor, a new original content contribution is recorded and other linguistic versions of the page get updated with the information.

The "Page Translation" box provides links for translators to view the relevant changes made to the page. When using those links, the translator is brought to a slightly different version of the edit page. The page displays the changes to be translated along with the text area. When the translator indicates that the translation of the changes is completed, the translation target gets marked as containing the changes provided by the translation source. Again, other linguistic versions of the page get updated with the new information.

The directed graph representation of the described scenario can be illustrated as in figure {architecture_graph.dot.png}. In the graph, white nodes are original content contributions and gray nodes are versions resulting from a translation efforts. Solid arcs are page evolutions from version to version and dashed arcs are translations from source to target. On each node, the original content contributions included in the version are listed on the second row.

The graph representation is in fact very close to the internal representation used. Beyond providing useful information for the site visitors and support translators, the entire translation history is preserved. Figure {en_history} presents the page history of the English page in the scenario. The information from the translation history will allow to analyze the translation patterns and evolution of the communities around the different linguistic versions of a page.

{img src=http://wiki-translation.com/tiki-download_wiki_attachment.php?attId=22 desc=en_history}


!!New scenario created by AD

!!!NOTE: I (AD) have BMP files for each step of the way on my laptop.

!!!CRAP: I created it on wiki-translation site which doesn't have the most recent version! It lacks the X% up to date messages! I will try to construct a story using the screen shots that LPH produced instead.

First, John Doe creates English page __Final Report__ in two consecutive edits. We now have:

  en_v2 = {e1, e2}
  fr = None
  es = None

Then, Marie Quidam sees the English page, and hits the __Traduire__ ("Translate") button. She enters French as the target language, and __Rapport Final__ as the name of the target page. The system pastes the content of the English page __Final Report__ into an edit box, and inserts a note "Under translation" at the top. Marie translates the first sentence, when the phone rings. She hits the __Traduction Partielle__ ("Partial Translation") button and answers the phone. The status of the various pages is now:

  en_v2 = {e1, e2}
  fr_v1 = {}
  es = None

The sytem shows the pages as not being up to date, because it says that the __en__ version is "better". When Marie comes back from her phone conversatio, she clicks on the update icon in front of __Final Report (en)__, translates the remaining two sentences, and this time hits the __Complete Translation__ button, to signal that she is done translating. Note however that Marie forgets to delete the __Under translation__ line at the top of the page.

We now have:

   en_v2 = {e1, e2}
   fr_v2 = {e1, e2}
   es = None

and the French version __Rapport Final__ is now displayed as being equivalent to the English __Final Report__. Later, Juan Del Pueblo sees the French page, hits the __???__ ("Translate") button and translates the whole content in one go, making sure to save using the __???__ ("Complete Translation") button. We now have:

   en_v2 = {e1, e2}
   fr_v2 = {e1, e2}
   es_v1 = {e1, e2}

and the Spanish page __Informe Final__ is listed as being equivalent to both the English and French versions. Now, Marie notices that she forgot to erase the __Under translation__ line at the top of the French page. So she edits it, erases the page, and saves using the __Modif. Mineure__ ("Minor Change") button. This means that the change is not considered as a change that needs translation. So the state of the three pages remains the same, and they are still displayed as being up to date with each other.

Next, John Doe adds three bullet points to the English version of the page. We now have:

  en_v3 = {e1, e2, e3}
  fr_v2 = {e1, e2}
  es_v1 = {e1, e2}

At this point, the French and Spanish versions are labeled as needing updating from English. But Marie is not in a translation mood and would rather add original content to the French version. No problem, she can do that, and she adds her own two bullet points to the list. We now have:

  en_v3 = {e1, e2, e3}
  fr_v2 = {e1, e2, e4}
  es_v1 = {e1, e2}

At this point, all three versions are labeled as missing some edits. Now that Marie has gotten her original edits out her mind, she is in the mood for translating, and decides to update from the English version (by clicking on the double green arrows in front of the English page name). The changes to be translated are highlighted in green at the top of the screen. Marie copies and pastes it into the French edit field, then translates it to French and saves as a complete translation. We now have:

  en_v3 = {e1, e2, e3}
  fr_v3 = {e1, e2, e3, e4}
  es_v1 = {e1, e2}

The French version is the most up to date, but English and Spanish still needs to be updated. Now, Juan wants to bring Spanish up to date. He can do this either by updating from English or French.

__What next... Maybe Marie can now incorporate John's changes, and later on, Juan can decide which of EN or FR to update from... Q: Does the tool help him make that decision?__

__NOTES FROM AD: Storyboard below is being replaced by new storyboard above__

!Other possible scenario based on keyframes from Rick Sapir's video

First, John Doe creates a page in English, in two consecutive edits.

__TO BE INSERTED HERE: Screen shot of the EN page__

   en_v2 = {e1, e2}
   es = None
   fr = None

Then, Juan Del Pueblo decides to translate it to Spanish. 

__TO BE INSERTED HERE: Screen shot of the dialog where he enters the language and name of target page__

__TO BE INSERTED HERE: Screen shot of the initial translation dialog, with the En content pasted into the ES page, with the note at the top saying it's under translation__

Note how the system automatically pasted the English content into the Spanish page, and inserted a noted at the top saying that it is under translation. Juan translatest the whole text in one go, and therefore erases the "Translation in Progress" notice. He saves the translation using the "Complete Translation" button. 

__TO BE INSERTED HERE: Screen shot of EN page, with 100% up to date notice, and a link to the EN__

We now have:

   en_v2 = {e1, e2}
   es_v1 = {e1, e2}
   fr = None

In other words, English and Spanish both contain the same edits, and there are no other linguistic versions that contain additional edits. This is reflected on the screen by the fact that the Spanish page lists the English page as an "equivalent translation", and it says that the Spanish version is "100% up to date".

Next, Marie Quidam sees the English page, and decides to translate it to French. She pushes on the "Translate" button, enters French as the language, and the name of the French page. But just as she is about to overwwrite the English content with its French translation, the phone rings. So she presses on the "Partial Translation" button.

__TO BE INSERTED HERE: Screen shot of FR page, with Spanish content in it and still the English content__ 

We now have:

   en_v2 = {e1, e2}
   es_v1 = {e1, e2}
   fr_v1 = {}

Note how the French version is still assumed to not contain any of the edits from the English, because Marie saved the page as a "Partial Translation" and the system has no way of knowing which of e1 and e2, if any, has been translated at this point. This is reflected on the screen since the French page lists both English and Spanish as being "better" translations. It also says that the French version is 66% up to date (note: we will explain later how this number is callculated and why in this case, it's not 0%). 

At this point, John Doe decides to add more content to the English page. This time though, the edit he does needs to be translated to all the other languages ASAP. So John checks the "Send urgent translation request" box before saving his edit.

__TO BE INSERTED HERE: Screen shot of FEN page being edited and about to be saved, with the "Send urgent translation request" check box selected__ 

We now have:

   en_v2 = {e1, e2, e3*}
   es_v1 = {e1, e2}
   fr_v1 = {}

where the asterisk beside e3 denotes that this edit needs to be urgently translated. At this point, Juan looks at the Spanish page and sees the urgent translation notice at the top, and decides to do something about it. He clicks on the double green arrows icon in front of the English page name. The system shows him the content of the urgent edit e3* at the top, along with an edit box at the bottom with the Spanish content. Juan copies and pastes the English e3* content into the Spanish page, and overwrites it with its Spanish translation. He saves it as a complete translation.

__TO BE INSERTED HERE: Screen shot of ES page without the Urgent translation request anymore__

We now have:

   en_v2 = {e1, e2, e3*}
   es_v1 = {e1, e2, e3*}
   fr_v1 = {}

In other words, Spanish is completely up to date, including the urgent edit, and this is reflected on the screen by the disappearance of the urgent translation notice.

Later still, Marie comes back from her phone conversation and decides to pickup the translation from English to French. 

__TO BE INSERTED HERE: Screen shot of French being translated from EN. The whole EN content is highligted (NOTE: IN RICK'S SNAPSHOT, ONLY PART OF THE EN CONTENT IS SELECTED. LPH THINKS IT'S BECAUSE RS MOCKED AROUND WITH PAGES IN WEIRD WAYS, TRANSLATING PAGES THAT ALREADY EXISTED, ETC... INITIALISATION OF HIS SITE WAS A BIT WEIRD IN OTHER WORDS. NORMALLY, THIS SHOULD NOT HAPPEN)__

Since Marie previously saved her translation as "partial", the system does not know which of e1 or e2 has been propagated to French. Therefore it highlights both of them (which in this case turns out to be the whole English content). Marie does more of the translation but still saves it as saves is as a "Complete" translation.

---

!From Context and Background section

In designing the Cross Lingual Wiki Engine project, the objectives were to improve the capacity to translate collaboratively and bring a true wiki experience to it. It had been identified that translation is a key issue in content democratization and that the solutions currently offered by wiki engines were simply insufficient. For content creators, it was duplication of efforts and for the site visitors, it was inaccuracy and uncertainty.

In our vision, it was possible for authors part of different linguistic communities to collaborate on content beyond the language barrier. To allow them to contribute in the language they are most comfortable in and not to fear to never be read. For the site visitors, it would allow them to get a clear picture of where the information is and allow them to reach it easily, without having to scan excessive amounts of text.

To reach a wide public, it was necessary to integrate the solution to an existing product. Developing a reliable, fully featured, wiki engine with multilingual effort that is ready for production use is a colossal effort. With so many open source implementations available and ready to receive contributions, to embark on a new effort was a complete waste of time.

Among the many engines available, one had to be selected. To reach global acceptance, WikiMedia, the engine behind Wikipedia, would have been the best choice by far. However, the large popularity of the engine forces its maintainers to be very conservative about the changes that are incorporated. It's very likely that the changes like those required by CLWE would have been left out, leaving availability to a small group of users using our patched version.

<SNIP>

However, the SUMO group has specific requirements that had to be met above the initial goals of content synchronization. The SUMO group requires very high quality content. For this reason, they use an approval workflow to make sure all content visible on the website is approved. Moreover, they needed a way to mark translations as potentially out of date to warn readers that significant changes were made on an other translation.
 
To categorize the different pages in the knowledge base, extensive use of tags is being made. In order to improve the experience of the readers, these tags also need to support multilingual correctly. However this aspect will not be covered in the current paper.
History

Enable pagination rows per page
Information	Version
Sun 20 of Apr, 2008 13:55 GMT anonymous	15	Actions View Source
Fri 18 of Apr, 2008 15:28 GMT lphuberdeau	14	Actions View Source
Fri 18 of Apr, 2008 15:03 GMT lphuberdeau	13	Actions View Source
Tue 15 of Apr, 2008 18:54 GMT alain_desilets	11	Actions View Source
Tue 15 of Apr, 2008 18:47 GMT alain_desilets	10	Actions View Source
Tue 15 of Apr, 2008 18:42 GMT alain_desilets	9	Actions View Source
Fri 11 of Apr, 2008 00:14 GMT alain_desilets	8	Actions View Source
Thu 10 of Apr, 2008 16:20 GMT alain_desilets	7	Actions View Source
Wed 09 of Apr, 2008 11:47 GMT alain_desilets	6	Actions View Source
Wed 09 of Apr, 2008 11:46 GMT alain_desilets	5	Actions View Source
Wed 09 of Apr, 2008 11:45 GMT alain_desilets	4	Actions View Source
Wed 09 of Apr, 2008 11:45 GMT alain_desilets	3	Actions View Source
Wed 09 of Apr, 2008 11:41 GMT alain_desilets	2	Actions View Source
Wed 09 of Apr, 2008 11:34 GMT system created from structure	1	Actions View Source