I built the ai-ticket-classifier repository on GitHub to automatically classify support tickets and suggest a response using sentence embeddings.
The project is written in Python.
Development started 2026-07, last updated 2026-07.
I use the all‑MiniLM‑L6‑v2 model from the sentence‑transformers library for embeddings.
Classification is performed via zero‑shot classification.
If a ticket doesn't clearly match any category well enough (below a similarity score of 0.35), instead of forcing a wrong guess, the system flags it for a human agent to review.