Research path

Research methods and scholar examples

Research training beyond introductory study: evidence synthesis, causal design, modelling, qualitative analysis and reproducible experiments.

Start with your research question, choose a method, study original materials and produce work others can examine. Sources are mainly in English, with bilingual reading tasks.

Learn through your research project

Find a question, read papers and learn with AI

Eight units with methods, worked examples and free primary sources. Practise here or offline, take work to your AI, then import feedback and revise.

Choose a research task

The workbooks and review criteria are platform-designed exercises for projects or group discussion, not assignments set by the source authors.

Scholar examples and professional resources

More filters

Clear filters

2 resources found

Reproducible experiments · Handbook and guideline · Advanced · subject foundations

NeurIPS Paper Checklist

NeurIPS · 官网现行页 / current official page

Reading target
Audit claims, limitations, reproducibility, error reporting and compute resources.
Prerequisites
Run a machine-learning baseline independently and understand data splits.
Platform-designed exercise
Reproduce a main table with configurations, seeds, cost, variability and discrepancies.
Method boundary
A conference checklist does not replace research-method training or a justified experimental design.
Access
Public guidance; verify the target year's rules when submitting.

Open workbook and review criteria →

Source checked 2026-10-25

Reproducible experiments · Handbook and guideline · Advanced · subject foundations

The Turing Way · Reproducible Research

The Turing Way Community · 持续维护 / living handbook

Reading target
Version control, data management, reproducible environments and code documentation.
Prerequisites
Basic scripting, command line and Git.
Platform-designed exercise
Rebuild a figure from raw data in a clean environment; have another person follow the README.
Method boundary
Rerunning a workflow does not establish that its data, design or measurements support the conclusion.
Access
Open online handbook.

Open workbook and review criteria →

Source checked 2026-10-25