Best Practices for Search in a Multi-Language and Multi-Geography Setup

Jayne Loh
Tera Contributor

We have a HR knowledge base setup to cater for employees across multiple geographies each with different language requirements.

 

We tested on using translation tasks and created multiple child versions of an article and found that the consumption of these articles are very low. We now have real-time language translation turned on so anyone can translate an article to their preferred language in a single click.

 

Challenge: While articles can be translated in real time, the article remains in source language (English) and it makes search impossible in this scenario.

 

Would like to know what are some implementation methods/options that we can consider? Thanks a lot in advance!

1 ACCEPTED SOLUTION

Per-Kristian BM
Kilo Guru

Hello

 

We had the same type of issue, we want articles to be search friendly in the native language the users wanted. We found that in order to do that, we needed the articles to be published and indexed, just like you did the first time.

 

Searchability comes more down to how you document the articles and your content standard. We use the KCS article template. In this template, there is an issue field, and a environment field. We renamed the environment to "related to" as it makes more sense to the user. We created a content standard that focus on being very precise on issue descriptions. We tested using an old article, which used to be known as "imoportant information about the new autosave feature in office365". no user have ever search for the phrase, hence no search results and waste of time translating. with a new content standard, we reduced it down the the just need for search and translations friendly language. example on the new article:

 

Title: Turn off autosave in Office 365

Issue:
-Do not want autosave

-turn off automatic saving

related to: 

-Office365

-powerpoint

-excel

Instructions:

whatever answer the user might need to this issue

 

when we wrote it this way, the translations came out the other end very well, and the issue and title closely matching the user way of describing their needs/issue, it matched what they were searching for. the automatic translations was also matching well in other languages as the descriptions was made in a very simple language using just short issue statements rather then complex language. 

 

we also this stopped spending time on manually doing the translations tasks, we ended up auto-publishing them as the quality was (not perfect) good enough for end-users. we instead branded the translated child articles as "this is a machine translated article, we value feedback should there be major translations errors"

 

View solution in original post

5 REPLIES 5

Jayne Loh
Tera Contributor

Thank you - it does seem that the only way now is to enable the multi-language search functionality on AI Search and have auto-translation enabled. To improve search, it is to include the relevant tags in the article body as the search tags do not get automatically translated.