Reproduce “Consistent yet Contextual: Learning Author-Level Mixtures of Style for Authorship Attribution” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Omkar Yadav, Michael Jiang, Kenan Alkiek, Jian Zhu, David Jurgens
The authors model each writer with recurring style mixtures for authorship attribution, rather than a single average. They fit Gaussian mixtures to document-style representations, choosing component counts on development data. On the 40,000-author SADIRI task, mixtures score 0.78 versus 0.76 on Success@8 and 0.64 versus 0.61 on mean P(author|doc), compared with centroids; the scores do not establish that components are pure stylistic traits.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.