Artifact Evaluation

About the Artifact Evaluation Process

This year, AIIDE will include an Artifact Evaluation, a chance for authors of accepted papers & posters to submit a companion software artifact with their paper. Example types of artifact include stand-alone software, web applications, datasets, plug-ins/extensions for existing tools, social media bots, and others.

The purposes of the artifact evaluation are:

  • To promote reproducibility of our research results by reviewing the claims made in the paper and how well they are supported by the corresponding software;
  • To promote reuse by encouraging authors to release software that is well-documented and easy to use by peers;
  • To recognize software artifacts as important scholarly contributions in addition to the paper itself.

All submitted artifacts will be stringently peer reviewed through a separate review process from the main conference. Artifact evaluation is an optional round of additional reviewing available to authors of accepted papers and posters; the quality of artifacts will not affect acceptance decisions for main conference papers in any way. One companion artifact may be submitted per paper.

FAQ: But don't we already have demos and Playable Experiences? Why introduce another thing?

The Artifact Evaluation is distinct from the Playable Experiences and Demos tracks in that the software artifacts are tightly coupled with a research result documented primarily in a conference paper. Passing the artifact evaluation process is a signal to other researchers in the field that your software is robust, reusable, and that your research claims are reflected in the artifact (thus promoting reproducibility).

To help authors distinguish between these submission types, we summarize their criteria below:

FAQ: Can I submit to artifact evaluation if I haven't submitted to the main conference?

No. Artifact evaluation is only conducted on artifacts associated with accepted papers in the main conference. Authors of accepted papers will be contacted regarding how to submit their artifact for evaluation after paper notifications are sent out.

FAQ: Do all accepted papers need to submit an associated artifact for evaluation?

No. Artifact evaluation is optional, though encouraged, for authors of accepted papers. Artifact evaluation follows a separate peer review process from papers, with a separate (though potentially overlapping) committee of reviewers from the main conference program committee. Artifacts that pass this review process will be publicized on the AIIDE website and announced at the main conference.

FAQ: Where can I learn more about artifact evaluation?

Artifact evaluation is a new process being adopted at conferences across computer science, and it is new for AIIDE in 2018. You can learn more about artifact evaluation here.


Feedback on this process is welcome, and questions can be directed to the artifact evaluation co-chairs or the program chair. More details regarding this process will be made available closer to artifact evaluation submission deadline.

Instructions for Artifact Submission

If you are the author of an accepted paper, please follow the guidelines below to submit your artifact. These guidelines come from the official AEC webpage.

Submission Process

Submissions via Easychair: https://easychair.org/conferences/?conf=aiide18

Create a new submission with the same title as your accepted paper. In the submission form, include two things:

  • the paper's abstract (to help with bidding)
  • a URL pointing to the artifact site

There is no “paper” to submit.

At the artifact site URL, give us access to:

  • the artifact
  • the accepted version of the paper
  • instructions

See below for additional details.

Artifact Submission

Irrespective of the nature of the artifacts, authors should create a single Web page (whether on their site or a third-party file upload service) that contains the artifact, the paper, and all necessary instructions.

For artifacts where this would be appropriate, it would be helpful to also provide a self-contained bundle (including instructions) as a single file (.tgzor .zip; please avoid exotic compressors) for convenient offline use: imagine the reviewer who wants to download a single file to expand and work with during a train or bus commute.

If it would be helfpul, please feel free to include a video that demonstrates the artifact running or explaining how it should be run.

The artifact submission thus consists of just the URL and any credentials required to access the files.

Confidentiality

We ask that, during the evaluation period, you not embed any analytics or other tracking in the Web site for the artifact or, if you cannot control this, that you not access this data. This is important for maintaining the confidentiality of reviewers. If for some reason you cannot comply with this, please notify the chairs immediately.

Code Artifacts Packaging

Authors should strongly consider one of the following methods to package the software components of their artifacts (though the AEC is open to other reasonable formats as well):

  1. A VM (Virtualbox/VMWare) image containing software application already setup in the intended run-time environment (e.g., Mobile phone emulator, Real time OS). This is the preferred option: It avoids installation and compatibility problems, and provides the best guarantees for reproducibility of the results by the committee. Authors using Linux might find the CDE tool useful for automatically creating a VM image from existing software without needing to manually re-install all dependencies within a VM. Another promising new system is ReproZip.
  2. A binary installable package. We invite the authors to use CDE (Linux) or MSI Installer (Windows) to package the binary application.
  3. A live instance running on the Web.
  4. A detailed screen-cast of the tool along with the results, especially if one of the following special cases applies:
    • the application needs proprietary/commercial software that is not easily available or cannot be distributed to the committee;
    • the application requires significant computation resources (e.g., > 24 hours of execution time to produce the results);
  5. An installation or update repository for the tool (e.g., Eclipse update site or a Debian APT repository). However, we urge the authors to test their repository on standard environments to avoid installation problems by the committee members.
  6. A “live notebook” that can provide a journal view of all the work done to create the results.
  7. An “executable paper” format that provides a full provenance of a paper's results. A promising recent effort in this direction is CodeLab.

In all cases, authors should make a genuine effort to not learn the identity of the reviewers. This may mean turning off “call home” features or analytics, or only using systems with high enough usage that AEC accesses will not stand out. In all cases where tracing is unavoidable, the authors should warn the reviewers in advance so the reviewers can try to take adequate safeguards.

Non-Code Artifacts Format

Non-code artifacts should preferably be in open document formats. For documents and reports, we invite the authors to use PDF, ODF, or RTF. We invite the authors to submit experimental data and results in CSV (preferred option), JSON, or XML file formats. In the special case that authors have to submit non-standard data formats, they should also provide suitable readers.