Strong reasons:
- Promise more than delivered (this is the first work….).
- Miss important references (closely related).
- Results are too incremental or too unconvincing, or lower than state-of-the-art.
- Poorly written or oraganized.
- Incorrect statements or statements without support.
- Lack of component analysis for important components.
Weak reasons:
- Novelty is minor or incremental. Just an extension of A, or similar to A, or combination of A and B.
- Motivation is unclear or arguable.
- No theoretical guarantee of the effectiveness.
- No technical contribution: too few math formulations or the proposed method is too straightforward.
- Formulations are too dense and hard to follow.
- Paper writing has some flaws (e.g., ambiguity, redundancy, a few typos or grammar mistakes).
- Improvement is not very significant.
- Lack of component analysis for less important components.
- Lack of qualitative analysis.
- Unfair or insufficient comparison with state-of-the-art. Miss some baselines. Sometimes we need to create baselines if necessary.
- Hyper-parameter analysis: too many hyper-parameters, unclear how to set hyper-parameter, sensitivity to hyper-parameters.
- No significant test.
- Miss some details (e.g., technical details or experimental details), not self-contained.
- Miss less important references.
- Miss analyses on time/memory/model complexity