Collection of tools and interfaces for analyzing Twitter/X data, community networks, and exploring LLM memorization of social media content.
Interactive tooling to test whether Grok-family language models can regurgitate archived tweets when prompted with partial text. Uses LLM memorization as a novel interface for extracting Twitter data from model weights rather than APIs.
Graph-based analysis of the TPOT (This Part of Twitter) community using the Community Archive (6.9M tweets, 12.3M likes, follow relationships). Includes data pipeline, network metrics computation, and interactive visualization for exploring status dynamics and community structure.
Each subdirectory is a self-contained project with its own dependencies, documentation, and tooling. See individual README files for setup and usage instructions.