Hi everyone,
I'm looking into Predictive Intelligence and classification models, and would like some advice.
Currently I've set up prediction on a couple of different fields, and it works fine.
Next up is to extend these predictions to run for a couple of different languages,
and I can't seem to find advice on how to best set this up.
My initial thought was to create different solution models,
one for each language (and use different stopwords for each model),
and run different solutions depending on the language settings on the caller of the record.
The models will be trained on records that are created by users with the same language setting.
The idea behind is to get more accurate solution models, with one model per language and field to predict.
The issue with this setup, is that it will require a large number of models, which are similar
in the config, only thing that varies is the language (and the data it's trained on). That feels wrong.
Is there another way this could be done?
Has anyone some advice to give in setting up prediction on the same field, but working in a multilingual setting?
