research-paper-writing

Follow the phased pipeline: define a one-sentence contribution, map each claim to a supporting experiment, run experiments with incremental result saving, then draft using the official venue LaTeX template. Finish with a self-review pass simulating reviewers and complete the venue's mandatory checklist before submitting.

Never generate BibTeX from memory; fetch it programmatically via DOI content negotiation from CrossRef. Verify each paper exists in at least two sources such as Semantic Scholar and arXiv, and mark anything unverifiable as [CITATION NEEDED] for manual review.

The templates directory includes official style files for ICML 2026, ICLR 2026, NeurIPS 2025, ACL, AAAI 2026, and COLM 2025, each with its .sty and .bst files plus example documents. Page limits range from 7 pages (AAAI) to 9 pages (ICLR, NeurIPS, COLM).

Yes, it covers annotator selection, pairwise versus Likert scale design, inter-annotator agreement metrics like Krippendorff's alpha and Cohen's kappa, crowdsourcing platform choices, and IRB guidance. It also lists the reporting details reviewers at ACL and EMNLP expect.

Use McNemar's test for paired binary outcomes between two methods, bootstrapped 95% confidence intervals for key metrics, and Cohen's h for effect size on proportions. Always report sample sizes, number of runs, and whether error bars show standard deviation or standard error.

Autoreason works best with mid-tier models on constrained tasks where the generation-evaluation gap is large. For frontier models on unconstrained tasks, use critique-and-revise or a single pass, since autoreason suffers synthesis drift without scope constraints.