Feature spotlight: Advancing language settings for content on LinkedIn
November 1, 2019
Co-authors: Nanna Ericson and Tetyana Bruevich
Over the last few years, LinkedIn has rolled out several features that provide instant translation of feed content, enabling members to experience content in their own interface language. The interface language is set by default to be the same as the browser language the member is using when signing up for LinkedIn, but it can also be changed in the LinkedIn Account Settings menu to any of the 24 languages offered.
We’ve now taken this personalized experience one step further by introducing two new settings that allow you to input all the languages you understand and do not want to be offered translations for, as well as being able to define which language you would like all content translated into. This makes the choice of content experience explicit, and also extends our translation offering beyond the LinkedIn interface languages to the 60+ offered for translation by Microsoft Cognitive Services.
In the scenario below, our member Maryia has met a colleague from Spain at a conference and would like to read her new connection’s post in the feed. Mariya is using LinkedIn in English, so the Spanish post is by default offered with a translation into English. But Mariya’s actual native language is Ukrainian, so her experience with the initial translation is less than ideal. As a result, she goes in to rate the translation, and is then offered the opportunity to modify her language preferences. Clicking “Language Settings” takes her directly to the site preferences page, where she can select Ukrainian as her preferred content language, and also add that she doesn’t need translations from her second language, Russian.
Maryia can now refresh her feed and read Savannah’s post in Ukrainian instead, along with any other posts in languages she has yet to master. She can also comment in her native language, and feel confident that Savannah and the rest of their network will be offered a translation into the language of their choosing as well.
The original translation rating module allowed members to send us feedback on the translation quality. This module has now been extended to include a direct entry point for adding your personal translation preferences, in addition to a standalone section on the general Account Settings page for adding and modifying your content language preferences.
Our language detection service analyzes all new content units and records the detected languages in a database. The service uses a combination of tools for content analysis, including multiple static libraries as well as Microsoft Text Analytics API. We also have locale mapping in place to standardize the language formats, since all tools and devices don’t adhere to the same locale formatting standard.
After analysis, we record the detected language(s) of each piece of content, along with a confidence score. This score allows us to identify multilingual posts, and to determine whether the content can still be translated. If the confidence score is high enough for one language, we can offer translation for the entire post or comment from the detected language into any language preferred by the member.
More translation options
Our original translation feature for feed posts used the LinkedIn interface language to determine the member’s preferred content language. This logic is persisted for any member who is happy with their current content language experience, but for those who prefer another language, or who speak additional languages to the one used for their interface, we now provide more customization. For these members, these explicit content language settings can save space in the feed by excluding the “See Translation” button from content they already understand, while also offering translations into more languages than previously offered.
We hope this provides an intuitive path for members of the global workforce to further customize their feed content experience, so please check out your language settings today.
Many thanks to the dedicated product folks, engineers, and designers who made this happen over the last couple of quarters, especially Clarisse Siu, Cissy Chen, Ivan Kirilov, Annie Lin, Nuo Dou, Angelika Clayton, Lena Berestov, Helen Ung, Jonathan Salvador, Xiang Lin, Lee Mallabone, Collin Yen, Dan Dancescu, Johnny Wang, Shipra Jain, Qianru Zhang, and Ricardo Rivera Ayala.