mediaite-ghostink
Deterministic scrape → extract → analyze → report — combining statistical stylometry, embedding drift, and optional probability and AI-baseline tracks.
What this is
Section titled “What this is”A deterministic four-stage forensic pipeline over a WordPress newsroom corpus. It produces statistical and documentary signals — change-points in stylometric features, embedding drift, convergence to an “AI baseline” — that reviewers can inspect, reproduce, and audit. The outputs are signals, not legal findings or definitive attribution of authorship or tool use.
Architecture Pipeline design, stage contracts, storage architecture, and data models.
Runbook Operational quick reference, debug commands, and environment setup.
CLI reference Auto-generated reference for every `forensics` command and subcommand.
Decision records Architecture decision records covering methodology, storage, and governance.
Forensic report
Section titled “Forensic report”Responsible use
Section titled “Responsible use”This project quantifies stylistic and structural shifts in published text. It does not identify the author of a piece, prove the use of a specific tool, or render a legal verdict. Reviewers should pair every reported signal with domain context, editorial records, and corroborating evidence before drawing conclusions.
Built by Abstract Data