Peer review is half of what this conference is about. Every submitting author reviews two submissions, and we also welcome volunteer reviewers — learning to evaluate scientific work, and to be evaluated, is part of the experience. These guidelines are inspired by the NeurIPS reviewer guidelines, adapted and simplified for a student conference.
Your two roles as a reviewer
A reviewer at this conference has two jobs, in this order:
-
Give constructive feedback. Your first goal is to help the authors make their work better. The review form has a dedicated section — “How I would make this submission better” — and it is the most valuable part of your review. Concrete, actionable suggestions beat general judgments.
-
Assess scientific quality. Your second goal is to give the committee an honest, argued evaluation of the submission, so that decisions and the conference program (posters and orals) are fair and well-informed.
Remember the spirit of the conference: it is pedagogical and inclusive. Most serious submissions will be accepted. You are not gatekeeping a prestigious venue — you are helping fellow students publish the best possible version of their work.
What a good review looks like
- Read the whole post carefully, and look at the project repository.
- Start with a short summary of the submission in your own words. It shows the authors you understood them (or reveals where the post is unclear).
- Be specific. “Section 3 lost me: define X before using it” helps; “the post is unclear” does not.
- Be kind and professional. Write as if discussing the work face-to-face with a colleague. Criticize the work, never the authors.
- Value the context. Submissions come from very different settings — an internship with supervision, a solo project done at home. Read the declared context of creation and evaluate the work for what it is.
- Say what you liked. Strengths matter as much as weaknesses, both for the authors’ morale and for the committee’s decision.
- Know the limits of your expertise, and say so honestly in the confidence score.
The review form
You will be asked for:
| Field | Content |
|---|---|
| Summary | The submission in your own words (a short paragraph). |
| Strengths | What works: ideas, execution, clarity, openness. |
| Weaknesses | What doesn’t, argued and specific. |
| How I would make this submission better | Your concrete suggestions — the heart of the review. |
| Quality (1–5) | Is the work well executed? Are the claims supported? |
| Clarity (1–5) | Is the post well written, well organized, accessible? |
| Correctness (1–5) | Are the technical content, experiments, and conclusions sound? |
| Context made explicit (0–1) | Does the post honestly explain the project’s story — where, why, and with whom it was done? |
| Overall recommendation | Accept / Reject, with a justification. |
| Confidence (1–5) | How sure you are of your assessment. |
Rules
- Double blind. You do not know who the authors are, and you must not try to find out. In particular, do not inspect GitHub accounts, commit history, or repository metadata to identify authors — we rely on good-faith anonymity.
- Conflicts of interest. Declare any conflict (you know whose project this is, you contributed to it, close friends…). When the reviewer pool allows it, we avoid same-school assignments.
- Confidentiality. Do not share or discuss submissions outside the review process until they are published.
- Where. All reviews and discussions happen on OpenReview — openreview.net/group?id=HPSCAI/2026/Conference — never in GitHub comments, which are reserved for technical rendering issues. Create your OpenReview account with your institutional email (automatic activation; personal emails can take up to two weeks of moderation).
Process and workload
- Each submission gets at least two reviewers; each reviewer handles two to three submissions at most.
- There is a single round: submission → reviews → decision → final version. No author rebuttal, so make your review self-contained and clear.
- Decisions (accept/reject, then posters and orals) are taken by the area chairs with the senior academic committee, based on your reviews. Awards are decided by the senior academic committee. See the committee page for the full process and conflict rules.
For any question: contact the Program Chairs (contact to be announced).