Repeatability Evaluation
Repeatability Evaluation
Authors of accepted Regular and Short papers are encouraged, but not required, to submit supporting artifacts for independent repeatability evaluation after paper notifications.
Eligibility and timing
Authors of accepted Regular and Short papers are eligible and encouraged, but not required, to participate in repeatability evaluation after paper notifications. Authors of tool-oriented Short papers are specifically encouraged to submit artifacts. Submission and notification dates remain tentative, and the separate artifact portal is not yet open. Badge terminology, final platform details, and exceptional arrangements will be announced before that process opens.
Artifact package
Submit one anonymized ZIP archive no larger than 600 MB. It must include an anonymized artifact-description PDF that identifies the included results, explains the system requirements, gives complete installation and execution instructions, describes the evaluation scenarios and expected outputs, and maps those outputs to the corresponding paper figures, tables, or claims. Explain any expected variation caused by stochastic behavior.
Include the software and accompanying data needed to reproduce the supported results, together with the complete implementation, dependencies and versions, a license, and clear setup and execution instructions. Automated installation and validation scripts are strongly encouraged. If the final instructions permit externally hosted material that cannot fit in the ZIP archive, every link and hosting account shown to evaluators must be anonymized and remain accessible and unchanged throughout evaluation.
Anonymity and portability
The complete artifact package must be anonymized: remove author names, affiliations, repository identities, account names, file metadata, and other identifying material from the archive, documentation, software, data, and any external links. Packages remain confidential review material; disable or clearly disclose analytics that could reveal evaluator identities.
A self-contained Docker image or OVA/OVF virtual machine is strongly encouraged. Software should run on a major operating system, or the archive should include the environment needed to run it. The package should run without network access when practicable; authors whose artifact requires licensed software, special hardware, unusually large data, or exceptional computing resources should follow the alternative-arrangement instructions in the final call. Packages that do not satisfy the final artifact instructions may be declined without evaluation.
