Page MenuHomePhabricator

[Epic] ContentTranslation - Lists of articles to translate
Open, HighPublic

Assigned To
None
Authored By
Amire80
Apr 15 2015, 3:20 PM
Referenced Files
F2661495: dashboard-flat-lists.png
Oct 6 2015, 8:45 AM
F2661501: birRrM2.png
Oct 6 2015, 8:45 AM
F177483: JSMi1YG.png
Jun 10 2015, 2:55 PM
F177485: WT7ESAT.png
Jun 10 2015, 2:55 PM
F177487: dbGY1Jn.png
Jun 10 2015, 2:55 PM
F177481: yuhQPbp.png
Jun 10 2015, 2:55 PM
F173690: dashboard-lists.png
Jun 3 2015, 10:52 AM
Tokens
"Love" token, awarded by Framawiki.

Description

Individuals and groups often follow some kind of list of articles to translate. Those lists are often informal (personal notes, a list in a user page or just an idea in the user's head) and lack integration with the rest of the translation process. By providing an integrated support for them, we can improve the whole translating experience.

This is a general Epic ticket to describe and discuss the idea. The workboard is at ContentTranslation-TranslationList. Specific tickets will be created for individual aspects. More details are provided below.

Goal

  • Increase engagement. Allow users to keep track of the articles they want to get translated easily. This will reduce missed opportunities for translation (e.g., not having time to translate the article they just found right now)
  • Increase discovery. Make it easy for users to find interesting articles to translate.
  • Increase collaboration. Sharing lists on specific topics that interest a community can encourage user participation. Participating together in a campaign can help to motivate users.
  • Track the progress of topic coverage in each language. Some languages cover some topics better than others, and this will help users understand which topics need better support and help complete it. The topics themselves can be defined by users and communities.

Scenarios considered

  1. Collecting articles for later. The user finds an interesting article but has no time to translate it now. The user can mark it so that it is easy to find later to do the translation. This can happen at different points:
    • Explicitly. The user just thought about an article and wants to add it for later.
    • While reading. Some content is missing or a page is not available in the user language (interlanguage link entry point).
    • When getting suggestions. The user finds an interesting suggestion, for later.
    • When translating. The user encounters missing links on articles that are worth creating in the user language.
    • When completing a translation. The user may want to add other articles from the same collection.
    • Based on other collections and the like. Other articles in the same category/collection/"list of"-article of a given article.
  2. Create a list of articles for a contest. The user is organising a campaign on their local wiki.
    • Add several articles efficiently.
    • Set the conditions of the campaign (languages, time period, etc.).
    • Allow other organisers to add/propose articles.
    • Share and connect the campaign with the information on-wiki.
    • Follow the progress of the campaign.
  3. Find interesting lists of articles.
    • Discover, keep accessing relevant lists (and not do so, when they are no longer relevant).
    • Get motivated on why pick a list and keep translating it.

Research information

There are many translation projects that use a task list that tracks the progress of translating a list of articles, with a different focus:

  • Some projects target single editors working on something that interests them. For example, with CX we already saw people translating articles on common topics, such as beaches of Spain, Fields medalists, or Queen albums. They can just look at categories in other languages that they know and see what's missing, or prepare a list of red links on their user pages.
  • Some projects target small groups of wiki editors working together. For example, in the Hebrew Wikipedia there is a small group of people that write or translate articles about UNESCO World Heritage sites. They have navigation templates for each country with the full list of heritage site in that country, with links to all the relevant future articles, and work according to these templates, gradually making all the links blue.

Learning the needs of these different groups of users and providing them with comfortable, automated and well-structured ways to create and update such task lists will help them focus on writing and translating and work less on the bureaucracy.

Solutions to explore

A prototype captures many of the design ideas which were later evaluated with user research.

yuhQPbp.png (1×2 px, 367 KB)

JSMi1YG.png (1×1 px, 132 KB)

WT7ESAT.png (1×1 px, 125 KB)

dashboard-flat-lists.png (1×1 px, 373 KB)

birRrM2.png (1×1 px, 342 KB)

  • Integrate the collections as part of the suggestions view. Suggestions, and "For later" will be just two special default collections.

List creation:

  • Allow to add articles to a list as an alternative to start translating for the different entry points:
    • Red links on a page (can be a recently translated page; a manually-created task list; a navigation template; etc)
    • Special pages (WantedPages, WithoutInterwiki)
    • Category in another language (Example: I know English and Hungarian. English Wikipedia has 200 articles about Nobel laureates in the appropriate category, Hungarian has only 70. I want to add the 130 missing articles to my task list in one click.)
  • Allow to add content not only from main namespace (e.g., allowing curated short article versions).

Campaigns and participation

  • Support list configuration and sharing
    • Support both monolingual ("I only care about translating from English to Hungarian") and multi-lingual ("I love Thai culture and I want articles about Thai music translated to as many languages as possible") cases for both the creators of campaigns and those accessing them.
  • Show user statistics and progress to motivate participation. (T114750)
  • Track progress of (big) campaigns you organised (presenting all the info organisers need about articles, participants and progress).

Discover lists and items to translate

  • Allow to create/add lists.
  • Surface shared lists when searching and getting suggestions.
  • Subscribe to get notifications for new opportunities to translate (e.g., subscribe/unsubscribe from lists to get notifications of new articles added).

The project is currently planned to be carried in three milestones:

  1. T114744: CXTL-MS1: Create a new translation list - planned for Language-Engineering October-December 2016
  2. T127016: CXTL-MS2: Share a Content Translation task list
  3. T114750: CXTL-MS3: Track translation task list progress

Each of them has its own subtasks. All are essential to the project's goals.

  1. Related tickets
  2. T87439 - Suggest articles to translate from the dashboard (can reuse at least some of the UI; suggestions list can come from personal or common task lists)
  3. T100283 - Simplify article selection to translate.
  4. Gather - it has lists in the database, although it requires more research about how useful it is to combine them

Related Objects

StatusSubtypeAssignedTask
OpenNone
OpenNone
OpenNone
OpenNone
DuplicateNone
DuplicateNone
OpenNone
ResolvedAmire80
OpenNone
OpenNone
Resolved santhosh
OpenNone
Resolved Petar.petkovic
OpenNone
Resolved santhosh
ResolvedPginer-WMF
Resolved santhosh
Resolved santhosh
Resolved santhosh
DeclinedPginer-WMF
ResolvedPginer-WMF
ResolvedAmire80
Resolved santhosh
ResolvedPginer-WMF
ResolvedPginer-WMF
OpenBUG REPORTNone
OpenNone
Resolvedngkountas
OpenNone
OpenNone
DeclinedNone
DeclinedNone
DeclinedNone
DeclinedNone
DeclinedNone
InvalidPginer-WMF
Resolved santhosh
OpenNone
OpenNone
OpenNone
Resolved Capt_Swing
OpenNone
Duplicate Capt_Swing
OpenNone
ResolvedPginer-WMF
OpenNone
OpenNone
OpenNone
OpenNone
OpenNone
OpenNone
OpenNone
OpenNone
OpenNone
OpenNone
ResolvedAmire80
OpenNone
OpenNone
Resolved santhosh
ResolvedKartikMistry
ResolvedKartikMistry
OpenNone
OpenNone
Resolved Petar.petkovic
OpenNone
OpenNone
Resolved Petar.petkovic
OpenNone
OpenNone
ResolvedPginer-WMF
ResolvedPginer-WMF
OpenFeatureNone
DeclinedNone
OpenNone
OpenNone

Event Timeline

There are a very large number of changes, so older changes are hidden. Show Older Changes

Yesterday Magnus was able to include a geolocation feature in his "Not in the other language" tool https://tools.wmflabs.org/not-in-the-other-language/? - meaning that it's now possible to find candidates for translation about things that are "near me".
To extend this concept... can I please suggest adding a "map view" to the translation suggestions?

If you could have an open-street-map inferface, move-and-zoomable to the place of your interest, with markers for all geolocated articles that exist in the source language but not in the target language, that would be a super-engaging way for people (as individuals, schoolgroups or at editathons) to chose articles.

specific example (as a kind of user story) I wrote this article about a street in Mexico in the english wikipedia https://en.wikipedia.org/wiki/Francisco_I._Madero_Avenue initially translated from the spanish wikipedia article - because I wanted to write about something nearby to where Wikimania will be this year. But, there was no easy way for me to identify this as an possible translation candidate - I had to click around randomly for a while and ask locals to tell me what was missing.

After talking with @AndyRussG at Wikimania, it would be interesting to take a look to Editor campaigns (and the technical design) for their usecases and interesting ideas on a related area.

This existing tool by Magnus Manske has some of the functionality you may be looking for

https://tools.wmflabs.org/listeria/dynamic.html#list=101

and with some small changes could do even more

https://bitbucket.org/magnusmanske/listeria/issues/11/create-links-for-create-article-and

Hi there, I want to bring a possible solution of this task from @Lucia Huang, a software engineer from Taiwan. What is her proposal please refer here for further details.

Thanks Shangkuanlc for introduce me here.

Hi all:

The idea is to use program to create a list of pages that requires translation and then prioritize the list with visiting number or other factors.

The motivation is from the time when I did TED video subtitle translation recently. I noticed there are many new volunteers joined the group each week and almost all the new video got assigned to someone soon after the video being put on-line.
And I noticed many interesting or important wikipedia pages do not have corresponding information in the Chinese page.
What if we can make the Wikipedia translation work similar to subtitle translations:
a. Prioritized list of translation tasks
b. Cut tasks into small items that each item can be finished in 4 hours by an experienced translator
c. Review policy
d. Credits for volunteers

I put the rough idea for #a at my personal page.

It is to figure out pages need better translation by some rules:

number_of_lines(target language page) / number_of_lines(English or original language page) < 10%

And prioritize the pages with factors including

  • recent visiting number of target language page
  • recent visiting number of English or original language page
  • recent edit times of English or original language page

Translation efforts on those tasks could give back the biggest influence and motivate volunteers to take the tasks.

Agree prioritizing by number of EN pageviews is an excellent idea.

Thanks for the input @Lucia_Huang. The prioritisation aspects are really relevant when exposing campaigns for users to translate. Since we expect many different campaigns to be created.

Currently the tool provides individual article recommendations and the number of views for articles are already considered. That work was done by the research team, and there are more details in this paper, and in T111028.

Amire80 updated the task description. (Show Details)

This is certainly a sensible usage scenario for this feature. It should make TOTW easier to share and use, and consequently more popular.

@Amire80 I am currently doing list-building research with @SGill around how to best build lists for Campaigns -- there is a larger need in the movement to have what I would describe as a "reusable worklist". Something that can be generated in both manual and semi-automated list building (i.e. we prototyped something like this at T187305 and https://tools.wmflabs.org/worklist-tool) -- that can then be integrated into both on-wiki and offwiki pages, and be portable so that if you want to use the same list for Event Metrics Education-Program-Dashboard or something like ListeriaBot, where its printed out into other pages.

The Worklist is a prerequisite to any collaborative effrot within the movement, and we should be able to generate them pretty easily with the various kinds of queries you can do with something like Wikidata or Petscan. Figuring out the portable format for that would be awesome :)

Whatever works. Translation is a major scenario for worklist / tasklist / campaign / whatever-you-call-it, but certainly not the only one.