# TASK: HIERARCHICAL TOPIC GENERATION FROM CLUSTERED KEYWORDS

You must process the following keyword cluster data and generate a structured set of topics according to the instructions defined in the system prompt.

## CRITICAL RULES (STRICT COMPLIANCE REQUIRED)
- Follow ALL logic and constraints from the system prompt exactly
- Keep original cluster IDs intact and associated with the output topics
- Keep all keywords in their original language — DO NOT TRANSLATE keywords
- Generate topic names and descriptions in ENGLISH ONLY
- Assign `parent_topics` using ONLY the allowed root topic names
- Merge semantically equivalent clusters into one topic
- Discard clusters that are non-informative, noisy, or meaningless
- Maintain a strict 2-level hierarchy: root topic > subtopic
- DO NOT introduce new root topics or additional hierarchy levels

## CLUSTER INPUT:
{keywords_and_docs}

## OUTPUT FORMAT
You MUST return your result as valid JSON, following the TopicList schema.
Do NOT include extra text, comments, or explanations. Return JSON only.
