Hjälp:Innehållsöversättning/Översättning/Inledande maskinöversättning
 |
OBS: När du redigerar denna sida samtycker du till att släppa ditt bidrag under CC0. Se hjälpsidorna för Public Domain för mer information. | |
När du lägger till en ny paragraf till översättningen kan du börja från början eller använda en automatiskt genererad översättning som inledning. När det går används en standard maskinöversättning som en inledande översättning. De olika alternativen, detaljerna om deras tillgänglighet, och övervägandena när man använder maskinöversättning beskrivs nedan.
Alternativ för inledande översättning
Med alternativet "inledande översättning" i verktygskolumnen kan man bestämma det inledande innehåll som ska användas som en startpunkt för varje paragraf. Tillgängliga alternativ är följande:
- Använd en maskinöversättningstjänst. Detta gör att du kan börja med en automatiskt översatt version av den ursprungliga paragrafen. Antalet och namnet på dessa alternativ kan variera. Alternativ såsom "Använd Apertium" eller "Använd Yandex" kommer vara tillgänglig beroende på vilka språk som stöds för dessa tjänster (mer i denna och nästa avsnitt).
- Kopiera det ursprungliga innehållet. Den ursprungliga paragrafen kommer kopieras över till översättningen. Trots att innehållet förblir på ursprungsspråket anpassas vissa element till målwikin. till exempel så kommer länkar peka till motsvarande artikel på målspråket och mallar kommer omvandlas till motsvarande. Översättare kommer behöver fortfarande skriva om innehållet fullständigt men de anpassade elementen kan vara enklare att återanvända.
- Börja med en tom paragraf. Att börja med en tom paragraf kan vara användbart då det alternativa innehållet kräver mer arbete än att bara skriva den.
Man kan snabbt byta mellan de olika tillvägagångssätten oberoende på varje paragraf då varenda en kan fungera bäst för olika sorters innehåll. Att byta mellan de olika tillvägagångssätten sparar ändringarna du gjort på paragrafen. In this way, you can try a different approach even if you started editing the original one without the fear to lose your changes if you finally decide to go back to the original approach. Two additional options are relevant in this context:
- The reset translation option is available when you made modifications to the initial content provided.
It allows you to restore the initial content by discarding the changes you made.
- The mark as default option is available when you select an approach for a paragraph that is not the default.
It allows you to set the default approach for the next paragraphs that are added to the translation. This can be very convenient if you found that a particular translation service works generally better than the default one.
Machine translation availability
Content translation integrates several translation services, and each service supports a different set of languages. The services supported are listed below with a link to the list of languages they support:
- MinT (languages supported)
- Apertium (languages supported)
- LingoCloud (languages supported)
- Google Translate (languages supported)
- Yandex (languages supported)
- Elia (formerly known as Matxin) (languages supported)
The lists of languages above point to the configuration code to make sure that the information is in sync with the way the tool currently works. The lists show the language code for the source language at the initial indentation level and the codes of all the supported target languages below it.
You can also check the available services for a specific language pair using our APIs. In this example for the Spanish-Catalan pair, you can see that Apertium is the default service (first in the list) among the other services available.
Languages are enabled gradually, based on the observed results and community feedback. It is possible that machine translation for some languages has not been enabled yet, even if they are supported by the underlying services.
Improving existing translation services
When you publish a translation with Content Translation you are helping translation services improve. All the corrections you make to the initial machine translations are exposed through an API and data dumps and can be potentially used to improve existing services. In addition to that, some of the translation services listed above provide specific ways you can contribute to their projects.
MinT
The translations provided by MinT are based on translation models such as NLLB-200 and OpusMT. The training data compiled by the Opus project is resused by many translation models, and you can contribute to it in order to help improve the translation quality for your language.
OpusMT is an open source neural machine translation system that is trained with multilingual documents freely licensed and available online. This open corpus is used to train the translation system, and expanding the corpus will lead to better translations. The contents generated by using Content Translation are integrated automatically in the corpus, but you can contribute to expand the corpus further:
- Propose new sources to be integrated in the open corpus. You can contact Jörg Tiedemann to propose a new data source to expand the corpus.
- Submit documents directly (still a preliminary prototype).
You can upload translated documents in various formats including translated webpages to be incorporated to the corpus.
- Try language models locally with OPUS-CAT (available for Windows)
- Contribute to other projects that contribute their translations to this open corpus such as Tatoeba, an on-line open community collecting translation for sentences in many languages.
OpusMT is based on MarianNMT which is also an open source project. People with technical knowledge and interested in machine learning can also contribute to improve it.
Apertium
Apertium is an open source rule-based translation system. You can contribute to the project by encoding the language rules of your language. This process requires both linguistic and advanced technical knowledge, but you can get support from the Apertium team to expand the translation support for a new language pair.
Google Translate
Google Translate is not an open source project, but there are still ways for users to contribute back:
- Join the Google Translate Community to provide translations that help train their system.
- Report bugs when the translation system shows unexpected behavior when dealing with certain elements such as spacing, numerals or end of sentence marks (view full list).
Expanding language support with new translation services
Content translation has been designed as an extensible platform. So it is possible to develop new clients to integrate additional translation services. Some considerations about the way translation services are integrated:
- Machine translations and the user corrections made are published as part of the data on published translations, which can provide a useful resource to create or improve your translation service.
- External services integrated only receive publicly available wiki content, and return a translated version of such content that is compatible with the licenses used in the wiki.
No personal information is shared with the translation services.
Feedback on the support provided for each language is very useful. Please, let us know if you are missing support for some language, or whether higher quality options are available for it. You can provide such feedback on the project talk page or in this ticket.
Considerations on machine translation
Machine translation is far from perfect when intended as a final outcome. However, many users find it very useful as a starting point. Please make sure to review the content from these different perspectives:
- Make sure the original meaning is preserved.
- Check that there is no information missing, especially for elements such as links, references and templates that include information that is not always visible on the surface.
- Read the translated content to make sure it reads naturally as an independent page.
Limitations with complex elements
In some cases the content may not appear in the translation as expected:
- Some of the services supported only work with plain text.
This means that formatting and rich content elements such as links and citations from the original article are lost in the translation, and Content translation needs to guess where those belong in the translated text. Re-adding those elements is not always perfect and some elements may be in the wrong position or applied to the wrong part of the text.
- Complex elements such as references or templates may use a different structure in each language, which makes it hard to transfer the content from one language into the other.
Make sure to review the contents inside those elements to make sure there is no important information missing.
Enforcing the review of machine translation
Several automatic mechanisms exist to enforce the review of the initial contents. In this way, the tool makes sure that the initial automatic translation is reviewed enough before the contents get published.