Reproduce “Towards Question-Answering as an Automatic Metric for Evaluating the Content Quality of a Summary” 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

Daniel Deutsch, Tania Bedrax-Weiss, Dan Roth

QAEval evaluates summaries by checking which facts in a human-written reference a candidate summary can answer questions about. It selects noun phrases, generates questions, answers them from the candidate, and scores the proportion answered correctly. On TAC’08, QAEval-EM’s system-level Kendall correlation with expert responsiveness was .76, versus .70 for Pyramid and .67 for ROUGE-2; individual-summary results were lower or competitive.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.