notebook-lm-research（notebook LM 研究）

NotebookLM Research Skill

Long-context document grounding and research synthesis capability powered by Google NotebookLM.

When to Use

Activate this skill when the task involves:

• Deep document analysis (PDFs, articles, reports)
• Multi-source research synthesis
• Citation extraction and verification
• Knowledge base building for content creation
• Literature review and summarization

Capabilities

1. Document Ingestion

Upload and process documents for analysis:

• Formats: PDF, Google Docs, web pages, text files
• Capacity: Up to 50 sources per notebook
• Context: 1M+ token window for comprehensive analysis

2. Research Synthesis

Extract and synthesize information:

• Key themes and patterns
• Contradictions and gaps
• Citation mapping
• Expert quotes and statistics

3. Query-Based Analysis

Answer specific research questions:

• Fact verification
• Comparative analysis
• Timeline construction
• Entity relationship mapping

Execution Pattern

1. INGEST → Add source documents to NotebookLM notebook
2. ANALYZE → Run initial summary and theme extraction
3. QUERY → Execute targeted research questions
4. SYNTHESIZE → Aggregate findings into structured output
5. CITE → Generate citation references for all claims

Output Format

Research outputs should follow this structure:

<research_output>
  <executive_summary>
    <!-- 2-3 paragraph overview -->
  </executive_summary>

  <key_findings>
    <finding source="[citation]" confidence="high|medium|low">
      <!-- Specific insight -->
    </finding>
  </key_findings>

  <themes>
    <theme name="Theme Name">
      <description><!-- Pattern description --></description>
      <sources><!-- List of supporting sources --></sources>
    </theme>
  </themes>

  <citations>
    <citation id="1" source="..." page="..." quote="..." />
  </citations>
</research_output>

Integration Points

• Content Orchestrator: Primary consumer for content creation workflows
• Google Slides Storyboard: Feeds research into presentation narratives
• GEO Marketing Agent: Provides citation vectors for authority scoring

Best Practices

1. Source Quality: Prioritize authoritative sources (academic, official, expert)
2. Citation Precision: Always include page numbers and direct quotes
3. Bias Detection: Flag potential biases in source materials
4. Freshness: Note publication dates for time-sensitive topics

Error Handling

Cost Considerations

• Fuel Cost: 10-30 PTS per research session
• Optimization: Cache frequently accessed research for reuse