I’m using Rasa 2.8.15, Rasa SDK 2.8.3, and Rasa X 1.0.1 for building a chatbot. With the Rasa X on NLU inbox and NLU training data pages, I’m not sure I found a software bug or not. Sentences in the Thai language are rendered/visualized character by character with an HTML span tag. The results you see in the attached image are hard for developers and users to read long sentences. Could you clarify that this character-by-character rendering is a feature for some purposes, or it does not cover other languages like Thai with vowels?
1 Like
I have fixed this problem. The problem I found is about the Thai font rendering engine for macOS with any browser.
I found several ways to fix this problem. The first solution is changing a font-family (CSS) to support the Thai font before ‘sans-serif’ e.g. Tahoma, Arial Unicode MS, Sukhumvit Set. And the second solution is changing a browser’s sans-serif font (advanced setting) from ‘Helvetica’ (or whatever) to a supported Thai font e.g. Tahoma, Arial Unicode MS, Sukhumvit Set.
I hope Rasa developers would take the first solution or a better solution to fix this kind of problem in either the Thai language or similar languages.
please mark your above post as a solution for other.@jaypung just change the main topic category to rasa X only and you will be able to see the solution next to every post.
1 Like