CornellNLP/ConvoKit

Built on a scikit-learn-compatible pipeline architecture, ConvoKit provides specialized modules for linguistic coordination analysis, politeness strategy detection, conversational hypergraph extraction, and neural forecasting of conversation outcomes (e.g., derailment prediction). The toolkit implements research-backed NLP features including utterance-level likelihood modeling, speaker linguistic diversity measurement, and pivotal moment identification, all operating on a unified Corpus data structure that standardizes metadata across diverse conversation sources.

625 stars and 2,984 monthly downloads. Actively maintained with 5 commits in the last 30 days. Available on PyPI.