Context:
Consider the following {granularity}(s), which occur before the {granularity} to be analyzed:
{context_window}

Task:
Review the extracted causal relationships for the following {granularity}:
{item}

Previously extracted relationships:
{prior_relationships}

Instructions:
1.Review the extracted relationships against the text.
2. Keep relationships that are clearly supported by the text.
3. Revise wording where needed so that each cause and effect is clear, concise, and faithful to the text.
4. If two relationships are duplicates or substantially overlap, merge them.
5. If one relationship combines distinct causal claims, split it into separate relationships.
6. Drop relationships that are unsupported, incorrect, or redundant.
7. Add missing relationships only if they are clearly and directly supported by the text. Do not infer or speculate.
8. For each relationship, ensure the direction is correct (cause → effect).
9. Write cause and effect as short, meaningful phrases (preferably concise noun phrases), rather than full sentences.
10. Preserve important qualifiers such as timing, quantity, degree, and negation.
11. Do not introduce information not supported by the text.
12. Focus on relationships expressed in the target {granularity}. Use the context only to clarify meaning, not to introduce new relationships.
13. The number of relationships should generally be similar to the original list unless there is a clear reason to change it.
14. Return the corrected full set using this JSON schema:
{json_schema}
Return only valid JSON and no additional text.