You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It looks for now use tf-idf with english-only preprocessing rules.
So for now non-english tags looks dummy.
I know it is possible to pass precomputed tags.
But prefer instead preprocessing steps were no so strict to better support non-english texts out of the box.
Activity
acylam commentedon Aug 26, 2025
I am also looking for a solution to this. How to add Chinese tags?
acylam commentedon Aug 26, 2025
@fredhohman If not calculated automatically, is there a way to add precomputed labels to clusters? Looking for a way to support non-English.
donghaoren commentedon Aug 26, 2025
No, there isn't a a way to add pre-computed labels yet, but we are looking into that.
I think we might be able to use the new
Intl.SegmenterAPI (see here) to provide better labels for more languages without adding a ton of dependency.