Chandassu: Open Source Framework for Telugu Poetry

This talk explores how open-source computational methods can digitally preserve endangered cultural knowledge systems, using Telugu Chandassu (classical metrical poetry) as a current case study. I'll present our approach to developing the first comprehensive digital framework for analyzing Telugu prosodic patterns, which achieved 91.73% accuracy in automated pattern recognition. I'll discuss how we built a collaborative dataset of 4,651 annotated padyams and designed culturally informed algorithms that respect traditional literary standards while enabling computational analysis. Also, I will be walking through 'chandassu' Python library, which we have published.

This work exemplifies FOSS principles by making cultural preservation tools accessible, transparent, and community-driven. Attendees will learn how computational social science methodologies can be applied to their own cultural heritage projects, and how open-source development practices enable collective intelligence around literary traditions that might otherwise be lost.

The Key Takeaways from this talk are:

1. Open Source Cultural Preservation Frameworks: Discover replicable patterns for building digital humanities tools that can be adapted to other endangered literary traditions and cultural knowledge systems.

2. Democratizing Literary Analysis: See how computational tools can make classical literary education more accessible to new generations.

3. Practical NLP Techniques for Low-Resource Languages: Understand approaches for developing linguistic processing tools when working with languages that lack extensive computational resources.