ARTIFICIAL INTELLIGENCE AND NATURAL LANGUAGE CONFERENCE
Purpose of this document This document defines the review criteria for papers submitted to the AINL Conference. Its purpose is to ensure consistent, fair, and constructive evaluations across all submissions. Please read it in full before submitting your review. All reviews must be completed through the submission system. Each reviewer fills in three fields: (1) Main Idea, (2) Detailed Remarks, (3) Decision. Guidance for each field is provided in the sections below. |
Field 1 — Main Idea
The Main Idea field is a short, neutral summary of the paper as you understand it. This is not an evaluation, it is a comprehension check. It helps the authors verify that their message was received correctly, and helps the Program Chairs understand whether a reviewer has engaged with the work.
What to include
- A 3–5 sentence summary of the paper’s central contribution.
- The problem being addressed and the proposed approach.
- The primary findings or conclusions.
- Do not copy the abstract verbatim.
- Do not include praise or criticism in this field — save that for Detailed Remarks.
- Do not paraphrase missing sections you expected to find; summarise only what is present.
Field 2 — Detailed Remarks
The Detailed Remarks field is the core of your review. It must be structured, evidence-based, and constructive. A high-quality review helps authors improve their work regardless of the outcome. Reviews that are vague, dismissive, or unsupported will be returned to the reviewer for revision.
2.1 Summary Assessment
Briefly characterise the overall quality of the paper in 2–4 sentences before detailing specific points.
2.2 Strengths
List the genuine merits of the paper. Be specific, generic praise such as “interesting topic” is not useful. Where possible, reference specific sections, figures, or results.
Aspects to consider
- Novelty: Does the paper address an open problem or propose a new angle on an existing one?
- Technical soundness: Are the methods described correctly, reproducibly, and with appropriate baselines?
- Empirical validation: Are the experiments well-designed? Do the results support the claims?
- Clarity: Is the paper well-written, with clear figures, tables, and notation?
- Relevance: Is the contribution meaningful to the conference audience?
2.3 Weaknesses
List the substantive problems with the paper. Distinguish between major weaknesses (which affect core validity or contribution) and minor weaknesses (which affect quality but not fundamental correctness).
Major weaknesses example categories
- Unsubstantiated claims: A central claim is not supported by the experimental results presented.
- Methodological flaw: The evaluation setup has a significant bias or confound that invalidates key conclusions.
- Missing baselines: The paper omits obvious prior work or competitive baselines that would challenge the findings.
- Reproducibility: Critical hyperparameters, code, or data are absent with no justification.
- Incomplete ablation: The contribution of individual components is not individually evaluated.
- Limited scale: Results are reported on a small or unrepresentative sample.
- Presentation: Figures are unclear, notation is inconsistent, or writing is difficult to follow.
2.4 Justification of Decision
Explain, in 2–4 sentences, how your assessment of the strengths and weaknesses leads to your decision. This section should make the logical connection explicit — do not simply restate the decision label. If reviewers disagree, this section is used by the Program Chairs to arbitrate.
2.5 Recommendations for Improvement
Provide specific, actionable suggestions the authors could realistically address. Frame recommendations constructively. Even for papers you recommend rejecting, this section helps authors improve their work for future submissions.
Field 3 — Decision
Select one of the five decision options from the dropdown menu. Your decision must be consistent with the reasoning in your Detailed Remarks. The table below defines each option.
Decision | Guidance |
Strong Reject | Fundamental flaws in methodology, claims, or ethics; or the contribution is negligible and beyond repair. Reject without invitation to resubmit. |
Weak Reject | Notable weaknesses that significantly undermine the paper, but the core idea may have merit. Requires substantial revision. Lean toward rejection. |
Ambivalent | The paper has both merits and shortcomings of roughly equal weight. Outcome depends heavily on other reviewers. |
Weak Accept | The contribution is valid and the paper is mostly sound. Minor revisions needed but likely acceptable with small improvements. |
Strong Accept | Excellent contribution. Clear novelty, rigorous methodology, and well-written. Recommend acceptance with minimal or no changes. |
Calibration guidance Reserve Strong Accept and Strong Reject for papers where you are confident in your assessment. If you have significant doubts about the work’s correctness or your own expertise in a sub-area, use Ambivalent and flag this to the Program Chairs. A well-reasoned Weak Accept or Weak Reject is more useful than an unsubstantiated Strong decision. |