Main Page

From MediaWikiExt

Jump to: navigation, search

Contents

Language Grid Extension

Language Grid Extention for MediaWiki

Language Grid is an Internet-based multilingual service infrastructure using various language resources of machine translators and multilingual dictionaries, developed by research institutes, which can be utilized as Web services.

Language Grid Extension enables users to select and combine machine translators and multilingual dictionaries registered into the Language Grid as an extension to MediaWiki. For example, it enables users to improve the translation quality by combining multilingual dictionaries created, such as Wikipedia dictionaries.

MediaWiki Extension applies the API provided by the Language Grid Extension. That is to say, the functions listed become available by installing the Language Grid Extension.

  • Service Settings
Settings of user interface for machine translators and multilingual dictionaries from the Language Grid Extension are provided. It is possible to setup multilingual dictionaries categorized by machine translators and also select machine translators from alternative candidates.
Ghibli's Demo
  • PageDict
    • PageDict enables users to add technical terms or community based words which appear in MediaWiki articles into the multilingual dictionaries categorized by article. PageDict is accessible for every user.
    • When translation is carried through an API, provided by the Language Grid Extension, it is possible to combine the PageDict with machine translators. The translation quality can be improved with the help of words registered in the PageDict.
    • Example
      • Original text in Japanese
      スタジオジブリの考え方はよく分からないですけれども,確かめるのはあまり意味のないことだと思います。宮崎アニメ「紅の豚」の中の豚になった人間や,「となりのトトロ」の中の大きなネコバスやススワタリ(まっくろくろすけ)を見ると,ジブリ作品では現実と空想の境界は明確ではないと思います。
      (Although I cannot understand Studio Ghibli's way of thinking very well, I think that is not that big of a deal. When we look at their anime movies, such as Porco Rosso where a man is cursed and transformed into a pig, or My Neighbor Totoro with the huge Cat Bus and Susuwatari (Pitch-Black Blackie), we can see that there is no clear boundary between reality and fantasy in Studio Ghibli's movies.)
      • MT (Machine translation)
      A way of thinking of the studio yellowtail isn't understood well, but I think it's that it's so insignificant to make sure. When the man who became a pig, a big cat bus in "next TOTORO" and SUSUWATARI in the Miyazaki animation "red pig" (inkily, ROSU KE) are seen, I think the boundary of actuality and a vision isn't clear to the yellowtail work.
      • MT with Wikipedia Dictionary
      A way of thinking of Studio Ghibli isn't understood well, but I think it's that it's so insignificant to make sure. When the man who became a pig, a big cat bus in "My Neighbor Totoro" and Susuwatari in the Miyazaki animation "Porco Rosso" (inkily, ROSU KE) are seen, I think the boundary of actuality and a vision isn't clear to the yellowtail work.
      • MT with Wikipedia Dictionary & PageDict
      A way of thinking of Studio Ghibli isn't understood well, but I think it's that it's so insignificant to make sure. When the man who became a pig in the Miyazaki animation "Porco Rosso", a big cat bus in "My Neighbor Totoro" and Susuwatari (pitch-black blackie) are seen, I think the boundary of actuality and a vision isn't clear to a Ghibli work.

Document and Source/Binary Code

License

Language Grid Extension was licensed under the GPL. Language Grid for MediaWiki was licensed under the LGPL.

Authors of Language Grid Extension

Language Grid Extention is developed by the members of Language Grid Project, National Institute of Information and Communications Technology.

Multilingual Liquid Threads

Multilingual Liquid Threads is a multilingual version of Extension:LiquidThreads, developed for discussions in MediaWiki. Messages from users are translated by the Language Grid Extension and displayed in the Multilingual Liquid Threads. It also uses Extension:LanguageSelector.

Demo

The demos above support Arabic, Chinese, English, French, German, Indonesian, Italian, Japanese, Korean, Malay, Portuguese, Russian, Spanish, Thai, Vietnamese.

Source Code

Source code of Multilingual LiquidThreads

Functions

  • Multilingual expressions according to the language selection
    • Two ways are provided to browse messages in different languages.
    1. The result of machine translators will be directly displayed if it is saved in the MediaWiki database.
    2. If there is no existing data, the message will be translated by machine translation, then displayed and saved into the database.
    • If the original message is refined, the saved translation result will be updated.
    • Machine translation is designated in the Service Settings of the Language Grid Extension.
  • Language Selection for posting
    • A pulldown menu of available posting languages has been added into the New Post and Reply form. Users are required to select posting languages before they post. Besides this, users can learn what the original language of the posted message is, by looking at the bottom of the message.
  • Human correction of machine translation results
    • Human correction of machine translated articles and a history page for translated articles is possible.
  • Original language display of posted messages.
    • Users can view the original text of a translated message independently.
  • Meta-translation (Translation of "discussions about translation")
    • It frequently happens that translation results of discussions regarding translation make no sense at all. (Imagine how to machine translate the following sentence: Is it correct that "Good morning" in English is "おはよう" in Japanese?.) Under these circumstances, a specific part of the sentence should be excluded from machine translation, instead of valuable information which can help us understand the target word. It is possible to exclude the part from translation by surrounding it with <skip_translation> and </skip_translation>.

Use Case Scenario

Here are some examples of discussions that needed support from Multilingual LiquidThreads:

  • Real scale experiment
An experiment for evaluating the Language Grid Extension for Wikimedia and Multilingual LiquidThreads was conducted at the end of September, 2010. The experiment task was to translate an English Wikipedia article in the category of US National Parks into Japanese. In this experiment, three Japanese native speakers used the functions provided by the Language Grid Extension for translation tasks, and collaborated with two Americans, native English speakers, through the Multilingual LiquidThreads. The participants created 85 English-Japanese dictionary entries in the PageDict, and exchanged 273 messages through the Multilingual LiquidThreads regarding the translation task. ‪From the experiment results we concluded that users with disparate language abilities were able to effectively communicate and collaborate through the Multilingual LiquidThreads and successfully make use of the PageDict in their translation activities.‬
  • Discussions in the Wikipedia Translation Experiment
  • Dictionary entries in the Wikipedia Translation Experiment
  • Scenario 1: The discussion in an article’s Talk page
In the Talk page of the article “Ponyo” in English Wikipedia, there is an ongoing discussion about “Ponyo is not a goldfish”. The argument is about whether “Ponyo”, the main character of a Japanese animation “Ponyo on a Cliff by the Sea”, is a goldfish or not. First, an English speaker asserts that “Ponyo is not a goldfish” based on the mannerism in the movie. A Japanese speaker with an opposing view then quotes an interview with the director Miyazaki Hayao, shown in Japanese, making it difficult for English speakers to understand. Furthermore, it’s not the only case in which Japanese contents are quoted making it hard for English speakers to catch up on the discussions.
  • Before applying Multilingual LiquidThreads
  • After applying Multilingual LiquidThreads (We picked up first few messages of the original posts and put them into Multilingual LiquidThreads as a demonstration. You can change languages, but you cannot post messages. For posting messages, please go to demo page.)
  • Scenario 2: Discussion in Wikimedia Commons
Most of the content in Wikimedia Commons ("Commons" for short) are language independent, such as photos. Therefore, the contents are referred from many articles in various language versions of Wikipedia. However, the English Village Pump is particularly more active than in other languages, making it inconvenient for non-English speakers.
For example, there was a case of a Japanese user who wanted to raise a Deletion request, in consideration of an inappropriate photo in Commons. Because of the rule regarding raising a Deletion request in English, it can take up to two weeks before a user, not good at English, turns to other Japanese users for help in the Japanese Village Pump("Ask for help about how to post Deletion request" in English (Japanese Village Pump)). Afterwards, volunteers may give advice on how to write English articles and how to post messages in the English Wikipedia, based on which other users can successfully submit messages in English (Deletion request about the picture of Omiya Station(English)). 4 users contribute 9 messages to this Japanese thread.

Authors of Multilingual Liquid Threads

Multilingual LiquidThreads is developed by the members of Ishida & Matsubara Laboratory, Graduate School of Informatics, Kyoto University.

Future works

The following functions will be necessary, provided there are requests from the Wikimedia community.

  • Revision history review of PageDict edits
    • Reviewing of revision history should be accessible.
  • Automatic Filtering of PageDict Entries
    • Along with the realization of import/export function, PageDict entries will be filtered regularly to eliminate the ones which are unrelated with current page and discussion with the purpose of getting rid of entry disorder.
  • Import/Export translation settings
    • Settings of translation path will be able to be exported from current page and imported into other pages.
  • Import external dictionary
    • Language Grid has enrolled many kinds of multilingual dictionaries, which can be included in translation paths. The PageDict of Wikipedia articles can also be shared with each other.
    • Furthermore, it should be possible to import multilingual dictionaries which utilize interlanguage links in Wikipedia.

Contributors

Toru Ishida,Masanobu Ishimatsu, Masahiko Shimada, Xia Linsi, Jun Matsuno, Akihiko Satoda,Noriyuki Ishida, Toshiyuki Takasaki

Contact

  • multilingual-mediawiki[at]langrid.org
(Please replace "[at]" with "@")
Personal tools