Skill UI

This comprehensive guide outlines the rigorous standards for designing, auditing, and reporting Natural Language Processing (NLP) experiments, specifically tailored for top-tier academic conferences like EMNLP. It covers critical areas including baseline fairness checks, data contamination audits, statistical significance testing, prompt sensitivity analysis, and structured human evaluation protocols, ensuring reported claims are robust, reproducible, and methodologically sound.

A comprehensive guide for auditing and improving the scientific rigor of hardware architectural evaluations (e.g., HPCA style). It emphasizes establishing a clear 'fidelity contract' defining tools, scope, and configurations. It requires that all performance claims are backed by credible instruments, mandates capturing full machine state provenance, ensuring honest baseline tuning, and reporting results via per-workload distributions rather than single means.

A comprehensive guide for researchers in Human Relations to structure and report analyses, ensuring the direct, auditable link between raw evidence and theoretical claims. It covers best practices for qualitative coding (Gioia-style), critical interpretation, and advanced quantitative modeling, ensuring the final contribution is substantive and transparent.

This guide provides comprehensive best practices for designing and auditing empirical experiments for top-tier ML conferences. It covers necessary ablations, robustness checks (e.g., data shift, seeds), fair baselines, and detailed compute reporting to ensure that reported gains are scientifically genuine, reproducible, and address common reviewer skepticism.

This guide provides a comprehensive framework for stress-testing Machine Learning experimental evidence before submission or rebuttal. It covers critical areas such as establishing strong baselines, conducting mechanism-isolating ablations, reporting variance, checking for data leakage, and disclosing compute costs. Use this to ensure your ML claims are scientifically sound, robust, and fair against rigorous academic scrutiny.

This guide outlines the rigorous standards for achieving robust reproducibility in robotics research papers, especially those involving physical hardware. It details a two-tier approach, defining results as either fully *rerunnable* (code/simulation) or fully *auditable* (hardware logs/protocols). Key requirements include creating comprehensive hardware specification ledgers, logging detailed sensor data (rosbags), separating deterministic from stochastic claims, and structuring the entire repository for peer review.

This guide provides rigorous standards for designing, validating, and reporting security research, particularly for high-stakes academic conferences like IEEE S&P. It covers critical areas including adaptive adversary modeling, measurement validity (sampling, ethics), statistical reporting, establishing strong baselines/ablations, and adhering to ethical constraints (IRB, authorization) to ensure evidence is robust and reproducible.

This guide provides comprehensive standards for designing, auditing, and documenting AI experiments for major conferences like IJCAI. It covers establishing strong baselines, conducting ablation studies, and reporting meticulous details on hyperparameters, computational infrastructure, and ethical risks, ensuring maximum reproducibility and academic credibility for AI research.

This framework provides a comprehensive methodology for designing, auditing, and reporting empirical measurements about large-scale systems like the Internet. It guides users to ensure measurements are scientifically rigorous, by matching claims to evidence, quantifying coverage bias, designing for temporal changes, and adhering to strict ethical guidelines and data provenance standards.

A comprehensive guide for designing, auditing, and reporting reliable robotic experiments, particularly for major conferences like IROS. It moves beyond simple success rates, detailing how to map every claim to verifiable evidence (trials, simulations, field data). Key areas covered include defining pre-set success criteria, reporting failure taxonomies, ensuring fair baselines, and quantifying the critical sim-to-real gap using advanced small-n statistical methods.

This guide establishes standards for reporting reproducibility in robotics research, addressing the difference between traditional ML and physical systems. It mandates that every claim must be classified as either 'rerunnable' (executable code/simulation) or 'auditable' (detailed logs, hardware specs, and protocols). It covers creating a complete evidence map, detailing the hardware specification ledger, explicitly reporting the sim-to-real gap, and writing an honest availability statement, all within the main paper body.

This comprehensive guide helps authors assemble and audit the Supporting Information (SI) for a Journal of the American Chemical Society (JACS) manuscript. It ensures every reported chemical result and structure is fully reproducible by an independent chemist, covering detailed procedures, full characterization data (NMR, HRMS, IR), stereochemistry, and crystallographic details (CIF/CCDC).