fascinating-indigo•2y ago
Chunks without accents
Hello, I'm importing content from a URL in the knowledge base but the accents don't appear in the chunks. IE: “Administração” shows “Administrao”. The system does not show ç or ã. Is there anything I can do?
3 Replies
In Voiceflow, the Knowledge Base (KB) parser service reads the KB document from storage and "chunks" the content using different techniques. This process involves converting each chunk into a vector (aka "embedding") that represents its "meaning". However, it seems like you're experiencing issues with special characters such as accents and diacritics not being displayed correctly in the chunks.
This could be due to the encoding of the document you're importing. If the document isn't encoded in UTF-8, special characters might not be displayed correctly. You might want to check the encoding of your document and ensure it's in UTF-8 format. If it's not, you can usually convert it to UTF-8 using a text editor or a dedicated tool.
If the issue persists after ensuring your document is in UTF-8 format, it might be a more complex issue related to how the parser service handles special characters. In this case, it would be beneficial to bring this issue to the attention of the community members who might have encountered similar issues and could provide further insights.
Sources:
- Voiceflow Documentation
- Voiceflow Knowledge Base Guide
fascinating-indigoOP•2y ago
But it’s not a document, it’s a URL
KB documents can be files, text or URLs.
We’ve released a new update for our parser that might resolve the issue.