amazon-science/dstc12-controllable-conversational-theme-detection

Data & code for DSTC12, Controllable Conversational Theme Detection track

41
/ 100
Emerging

This project helps customer support, sales, and marketing teams quickly understand the main topics within customer conversations. It takes raw conversation transcripts and user preferences for grouping similar messages, then automatically generates clear, concise theme labels like "report defective product" or "schedule appointments." The goal is to provide a customizable summary of customer needs and inquiries.

Use this if you need to automatically categorize and summarize the key themes from a large volume of customer interactions, adapting the level of detail to your specific needs.

Not ideal if you primarily need to classify customer intent for an automated dialogue system with predefined responses, as this focuses on broader, user-driven themes.

customer-support conversation-analysis sales-enablement marketing-insights contact-center-optimization
No Package No Dependents
Maintenance 6 / 25
Adoption 5 / 25
Maturity 16 / 25
Community 14 / 25

How are scores calculated?

Stars

9

Forks

3

Language

Python

License

Last pushed

Nov 05, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/llm-tools/amazon-science/dstc12-controllable-conversational-theme-detection"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.