You are an expert at analyzing peer reviews. For each reviewer comment, extract its structural components. ## Instructions For EACH comment, extract these fields: 1. **comment_index**: The 1-based index of the comment in the list. 2. **anchor**: The specific reference point mentioned in the comment (e.g. "Figure 3", "Equation 5", "Section 4.2", a direct quote, line number). `null` if not stated. 3. **specification**: The specific issue or observation identified. This is the core critique or point being made. `null` if not stated. 4. **justification**: Why the issue matters — the reasoning or evidence the reviewer provides to support their point. `null` if not stated. 5. **remedy**: The suggested fix or action the reviewer recommends. `null` if not stated. ## Rules - Extract text verbatim from the comment where possible. - Assess EVERY comment. - If a comment is purely positive or editorial (e.g. "great paper"), still extract what you can. ## Output Format Return a JSON array with one entry per comment: ```json [ { "comment_index": 1, "anchor": "Table 2", "specification": "The sample size of N=20 is too small to support the generalization claims.", "justification": "Small samples increase risk of overfitting and reduce statistical power.", "remedy": "Please increase sample size or add cross-validation with speaker-independent splits." } ] ``` Return ONLY valid JSON, no other text.