Artifact Evaluation (AE) for RTAS is an optional evaluation process for research works that have been accepted for publication at RTAS. Due to the systems-oriented scope of RTAS, authors are strongly encouraged to submit artifacts to increase the value and the visibility of their research.
The AE process seeks to further the goal of reproducible science. It offers authors the opportunity to highlight the reproducibility of their results, and to obtain a validation given by the community for the experiments and data reported in their paper. In the AE process, peer practitioners from the community will follow the instructions included in the artifacts and give feedback to the authors, while keeping papers and artifacts confidential and under the control of the authors.
The AE process is non-competitive and success-oriented. The acceptance of the papers has already been decided before the AE process starts and the hope is that all submitted artifacts will pass the evaluation criteria. Authors of artifacts are expected to interact with the evaluators to fix any possible technical issues that may emerge during the review process, and to improve the portability of the artifact.
Authors of papers corresponding to artifacts that pass the evaluation will be entitled to include, in the camera-ready version of their paper, an RTAS AE seal that indicates that the artifact has passed the repeatability test. Authors are also entitled to, and indeed encouraged, to also use this RTAS AE seal on the title slide of the corresponding presentation at RTAS’19.
Recommendation: Use Virtual Machines
Based on previous experience, the biggest hurdle to successful reproducibility is the setup and installation of the necessary libraries and dependencies. Authors are therefore encouraged to prepare a virtual machine (VM) image including their artifact (if possible) and to make it available via HTTP throughout the evaluation process (and, ideally, afterwards). As the basis of the VM image, please choose commonly-used OS versions that have been tested with the virtual machine software and that evaluators are likely to be accustomed to. We encourage authors to use VirtualBox (https://www.virtualbox.org) and save the VM image as an Open Virtual Appliance (OVA) file. To facilitate the preparation of the VM, we suggest to use the VM images available at https://www.osboxes.org/.
The submission website is available to authors of accepted and shepherded papers at the following link: https://rtas19-ae.mpi-sws.org/.
When submitting an artifact for evaluation, please provide a document (e.g., a PDF, HTML, or text file) with instructions on how to use the artifact to reproduce the results in the paper. The document should include a link to the virtual machine image and a description of key configuration parameters for the virtual machine (RAM, number of cores, etc.), as well as the host platform on which you prepared and tested your VM image.
Please provide precise instructions on how to proceed after booting the image, including the instructions for running the artifact. Authors are strongly encouraged to prepare readable scripts to automatically launch the experiments. In the case the experimentations require a long time to complete, the authors may prepare simplified experiments (e.g., by reducing the number of samples over which the results are averaged) with shorter running times that demonstrate the same trends as observed in the complete experiments (and as reported in the paper).
Finally, be sure to include a version of the accepted paper related to the artifact that is as close as possible to the final camera-ready version.
A good “how-to” guide for preparing an effective artifact evaluation package is available online at http://bit.ly/HOWTO-AEC.
Artifacts for accepted papers are expected to be submitted on or before January 11, 2019 (three weeks after notification).
The evaluators may give early feedback if there are any issues with the artifacts that prevent them from being run correctly. Notification of the final results of artifact evaluation will be given about one week prior to the camera-ready deadline (by February 8th, 2019).
If you are not in a position to prepare the artifact as above, or if your artifact requires special libraries, commercial tools (e.g., MATLAB or specific toolboxes), or particular hardware, please contact the AE chair as soon as possible.
The artifact evaluation criteria are similar to those previously used by other conferences in their repeatability and AE processes. Submissions will be judged based on three criteria — coverage, instructions, and quality, as defined below — where each criteria is assessed on the following scale:
- significantly exceeds expectations (5),
- exceeds expectations (4),
- meets expectations (3),
- falls below expectations (2),
- missing or significantly falls below expectations (1).
In order to be judged “repeatable” an artifact must generally “meet expectations” (average score of 3 or more), and must not have any missing elements (no scores of 1). Each artifact is evaluated independently according to the listed objective criteria. The higher scores (“exceeds” or “significantly exceeds expectations”) in the criteria will be considered aspirational goals, not requirements for acceptance.
What fraction of the appropriate figures and tables are reproduced by the artifact? The focus is on figures or tables in the paper containing computationally generated or processed experimental evidence used to support the claims of the paper. Other figures and tables, such as illustrations or tables listing only parameter values, are not considered in this calculation.
Note that satisfying this criterion does not require that the corresponding figures or tables be recreated in exactly the same format as appears in the paper, merely that the data underlying those figures or tables be generated faithfully in a recognizable format.
A repeatable element is one for which the computation can be rerun by following the instructions provided with the artifact in a suitably equipped environment. An extensible element is one for which variations of the original computation can be run by modifying elements of the code and/or data. Consequently, necessary conditions for extensibility include that the modifiable elements be identified in the instructions or documentation, and that all source code must be available and/or involve calls to commonly available and trusted software (e.g.: Windows, Linux, C or Python standard libraries, Matlab, etc.).
The categories for this criterion are:
- None (missing / 1): There are no repeatable elements.
- Some (falls below expectations / 2): There is at least one repeatable element.
- Most (meets expectations / 3): The majority (at least half) of the elements are repeatable.
- All repeatable or most extensible (exceeds expectations / 4): All elements are repeatable or most are repeatable and easily modified. Note that if there is only one computational element and it is repeatable, then this score should be awarded.
- All extensible (significantly exceeds expectations / 5): All elements are repeatable and easily modified.
This criterion is focused on the instructions intended for other practitioners that seek to recreate the paper’s computationally generated results. The categories for this criterion are:
- None (missing / 1): No instructions were included in the artifact.
- Rudimentary (falls below expectations / 2): The instructions specify a script or command to run, but little else.
- Complete (meets expectations / 3): For every computational element that is repeatable, there is a specific instruction which explains how to repeat it. The environment under which the software was originally run is described.
- Comprehensive (exceeds expectations / 4): For every computational element that is repeatable there is a single command or clearly defined short series of steps which recreates that element almost exactly as it appears in the published paper (e.g.: file format, fonts, line styles, etc. might not be the same, but the content of the element is the same). In addition to identifying the specific environment under which the software was originally run, a broader class of environments is identified under which it could run.
- Outstanding (significantly exceeds expectations / 5): In addition to the criteria for a comprehensive set of instructions, explanations are provided of:
- all the major components / modules in the software,
- important design decisions made during implementation,
- how to modify / extend the software, and/or
- what environments / modifications would break the software.
This criterion explores the means provided to infer, show, or prove trustworthiness of the software and its results. While a set of scripts which exactly recreate, for example, the figures from the paper certainly aid in repeatability, without well-documented code it is hard to understand how the data in that figure was processed, without well-documented data it is hard to determine whether the input is correct, and without testing it is hard to determine whether the results can be trusted.
If there are tests in the artifact which are not included in the paper, they should at least be mentioned in the instructions document. Documentation of test details can be put into the instructions document or into a separate document in the artifact.
The categories for this criterion are:
- None (missing / 1): There is no evidence of software documentation or testing.
- Rudimentary documentation (falls below expectations / 2): The purpose of almost all files is documented (preferably within the file, but otherwise in the instructions or a separate README file).
- Comprehensive documentation (meets expectations / 3): The purpose of almost all files is documented. Within source code files, almost all classes, methods, attributes and variables are given lengthy clear names and/or documentation of their purpose. Within data files, the format and structure of the data is documented; for example, in comma separated value (CSV) files there is a header row and/or comments explaining the contents of each column.
- Comprehensive documentation and rudimentary testing (exceeds expectations / 4): In addition to the criteria for comprehensive documentation, there are identified test cases with known solutions which can be run to validate at least some components of the code.
- Comprehensive documentation and testing (significantly exceeds expectations / 5): In addition to the criteria for comprehensive documentation, there are clearly identified unit tests (preferably run within a unit test framework) which exercise a significant fraction of the smaller components of the code (individual functions and classes) and system-level tests which exercise a significant fraction of the full package. Unit tests are typically self-documenting, but the system level tests will require documentation of at least the source of the known solution.
In case of any questions or concerns, or for advice on how to best package and submit complex artifacts, please contact the Artifact Evaluation Chair.
Alessandro Biondi, Scuola Superiore Sant’Anna
Anway Mukherjee, Virginia Tech
Corey Tessler, Wayne State University
Daniel Casini, Scuola Superiore Sant’Anna
Gautham Nayak Seetanadi, Lund University
Martina Maggio, Lund University
Mohamed Hassan, University of Guelph
Paolo Pazzaglia, Scuola Superiore Sant’Anna
Yecheng Zhao, Virginia Tech