Systematic Literature Review and Synthesis

Objective: Create a structured, reusable AI prompt that guides large language models through systematic literature review and synthesis by extracting key claims, mapping supporting and opposing evidence against a research question, detecting contradictions, and producing a coherent narrative with explicit confidence levels for every conclusion.

To form an effective base prompt for this use case, follow these steps:

• Define the core research question and the set of papers or extracts the model will process so the AI stays focused on one clear inquiry.
• Instruct the model to extract only the most important claims from each source while recording the original citation or paper identifier.
• Require an explicit mapping of every claim as supporting, opposing, or neutral relative to the research question.
• Direct the model to surface any direct or indirect contradictions between sources and to explain the nature of each conflict.
• Ask for a final synthesized narrative that weaves the mapped evidence into a logical story and attaches a confidence level (high, moderate, or low) to each major conclusion based on the volume, consistency, and quality of the underlying evidence.
• Embed consistent style and output format rules so every run produces comparable, easy-to-review results.

Extended example values for the elements:

1. {{snippet:role_definition}}

   • You are an expert systematic review analyst skilled at rigorous evidence synthesis.
   • You are a senior research methodologist specializing in evidence-based literature synthesis for academic and policy audiences.

2. {{research_question}}

   • Does remote work improve long-term employee productivity in knowledge-based industries?
   • What is the effect of intermittent fasting on metabolic markers in adults with type 2 diabetes?

3. {{source_extracts}}

   • Paper A (2023): “Remote workers reported 12 % higher output after six months...”
   • Paper B (2024): “After controlling for selection bias, remote arrangements showed no sustained productivity gain...”
   • Study C (2022): “Self-reported focus scores declined by 8 % among fully remote teams after one year.”

4. {{snippet:extraction_instructions}}

   • Extract only the central empirical or theoretical claims from each extract. Record the source identifier with every claim.
   • Pull the primary findings and any stated limitations; ignore methodological detail unless it directly qualifies a claim.

5. {{snippet:mapping_instructions}}

   • Label each claim as Supporting, Opposing, or Neutral with respect to the research question and briefly note the strength of the evidence.
   • Classify every claim by direction (pro / con / mixed) and assign a provisional evidence weight (strong / moderate / weak).

6. {{snippet:contradiction_detection}}

   • Identify any claims that directly or indirectly conflict. State the nature of the contradiction and the sources involved.
   • Flag both statistical and conceptual disagreements and note whether they arise from differing populations, measures, or time frames.

7. {{snippet:synthesis_rules}}

   • Weave the mapped evidence into a coherent narrative that acknowledges both consensus and disagreement. Attach a confidence level to every major conclusion.
   • Produce a balanced synthesis that surfaces the strongest convergent findings first, then addresses residual uncertainty with explicit confidence ratings.

8. {{snippet:brand_voice}}

   • Maintain a precise, neutral academic register suitable for peer-reviewed reporting.
   • Write in clear, jargon-light language that remains fully rigorous and citation-ready.

9. {{tone}}

   • formal and objective
   • measured and evidence-focused