sanjanabhat846/code-mixed-emotion-intent

This repository is a data-engineering foundation for Indian code-mixed emotion, intent, and token-language research—not a trained classifier or inference service. Its main workflow ingests benchmark or volunteer records, builds canonical data, preprocesses utterances, validates and audits datasets, and prepares human-annotation batches. The README documents raw, interim, annotated, and processed tiers; the tree also contains components for annotation agreement and data-quality analysis. Some subsystem links below are documented only by module references in the SOURCE INDEX; unsampled internals are not assumed.