I’m currently working on an app for a video sharing website. I have at this stage manticore working with english language, however the app will support several languages. Each video has a title (original language) and a translated title for each language (example zh, ro, ru…probably all supported languages by google translate). What is the best approach to make this work? I have the following ideas:
One table for each language (video_id, title)
One table for all languages (different columns title_zh, title_ro, title_ru)
One table for all languages (one mixed column)
Are there any articles where i could read about this?
After doing some research i’ve decided to use method 2. One table for all languages, using separate columns for each translated title. I will try to identify language of search query and then search only in the corresponding language column. If anyone has experience with this, please let me know if there are better ways to do this! Thank you!
Are you planning to use different types of morphology for each language? If you already know the language before inserting a field value, and the list of languages is fixed, it makes sense to take advantage of that by using separate fields for each language.
Unfortunately, Manticore doesn’t support setting morphology per field - you can only apply multiple morphologies to the entire table. If that doesn’t work for you, another option is to have one table per language (option 1), and then combine them into a distributed table.
Yes, i know the language before inserting a field. I will have to test and do research regarding the morphology. I will try using morphology for each language. Will also try to use charsets for all languages. When a request comes, i will identify the language and only search in corresponding field (to make it faster). Its just a plan! I have to test everything now…
it is very simple to make it
for example
[{“id”:“1”,“model”:“12345”,“name”:{“en-gb”:“Sony VAIO”,“de-de”:“Sony VAIO 2”},“meta_title”:{“en-gb”:“Sony VAIO”,“de-de”:“Sony VAIO”}}]
after sending this array to manticore, you can switch or search the filed name by name.en-gb or name.de-de
that is it.