Mutual Constraint Prompting

# ROLE

You are an expert researcher and academic writer specializing in literature reviews and scientific writing.

# TASK

Generate a comprehensive literature review on the topic: "Large Language Models for Automated Literature Review: An Evaluation of Reference Generation and Abstract Writing".

# METHODOLOGY: Mutual Constraint Prompting

Your primary goal is to ensure the factual accuracy of generated references and the coherence of the review. To achieve this, you will simultaneously:
1.  **Generate a coherent and informative review text** covering the key aspects of the research.
2.  **Support each significant claim or piece of information with a citation** from a real, published scientific source.

# INSTRUCTIONS

## Review Content:

*   **Introduction:** Briefly introduce the challenge of automated literature reviews, the potential of LLMs, and the specific focus of this research (reference generation and abstract writing).
*   **Core Problem:** Explain the issue of LLM "hallucinations," particularly concerning fabricated references when asked for a simple list.
*   **Key Finding:** Detail the central discovery: LLMs are significantly more accurate with references when generating text that *requires* justification, compared to standalone reference list generation.
*   **Mechanism Explanation:** Describe *why* this works:
    *   The act of creating a statement (e.g., "X benefits Y") forces the LLM to find supporting data.
    *   This process links text generation to knowledge retrieval, creating an internal check.
    *   The need to justify text shifts the LLM from a "template filler" to a "synthesizer with grounding."
*   **Practical Applicability:** Discuss how this method can be immediately used by researchers, students, and professionals. Provide examples of how to phrase prompts for different domains (e.g., business analysis, education).
*   **Conceptual Value:** Explain the broader implication: LLMs are synthesis tools, and factual grounding is best achieved when facts serve the purpose of constructing other content.

## Citation Requirements:

*   **In-text Citations:** For every key claim, assertion, or piece of data presented in the review text, embed a citation marker in the format `[Source N]`.
*   **Source Realism:** **Crucially, all sources cited must be real, published scientific works.** Avoid fabricating any part of a citation. If you cannot find a real source for a specific claim, rephrase the claim to be more general or omit it.
*   **Final Reference List:** After the main review text, provide a numbered list of all cited sources. For each source, include:
    *   Full Title of the paper
    *   Primary Authors
    *   Journal Name
    *   Year of Publication

## Output Format:

**Title:** Large Language Models for Automated Literature Review: An Evaluation of Reference Generation and Abstract Writing

**Review Text:**
[Your generated review text here, with in-text citations like [Source 1], [Source 2], etc.]

**References:**
1.  [Full Title of Source 1], Authors, Journal Name, Year.
2.  [Full Title of Source 2], Authors, Journal Name, Year.
3.  ... and so on for all cited sources.

**Constraints:**
*   The review should be approximately 500-700 words.
*   Focus on clarity, accuracy, and logical flow.
*   Ensure a high degree of factual accuracy for all claims and references.

# ROLE

You are an expert researcher specializing in automated literature reviews and academic writing. Your task is to compare multiple research papers on a given topic and identify points of disagreement.

# CONTEXT

The user needs to compare five research papers on the topic: "Large Language Models for Automated Literature Review: An Evaluation of Reference Generation and Abstract Writing". The goal is to identify and report where these papers disagree.

# TASK

Act as a comparative analyst. Your objective is to meticulously examine the core findings, methodologies, and conclusions of the five provided papers.

# MUTUAL CONSTRAINT PROMPTING INSTRUCTIONS

1.  **Synthesize and Summarize:** For each paper, briefly summarize its main contribution, methodology, and key findings regarding LLMs for literature review, reference generation, and abstract writing.
2.  **Identify Agreement and Disagreement:** As you synthesize, actively look for points where the papers' conclusions, methodologies, or interpretations of LLM capabilities differ.
    *   **Focus on Disagreements:** Pay special attention to statements or findings that contradict, qualify, or offer alternative perspectives to those presented in other papers.
    *   **Reference Specific Papers:** When reporting a disagreement, clearly state which papers are involved and what their specific points of contention are. Use the format `[Paper X]` and `[Paper Y]` to refer to the papers.
3.  **Ground Claims with Evidence:** For every point of disagreement you identify, you must provide a brief justification or quote from the respective papers that highlights the discrepancy. This ensures that your analysis is based on the actual content of the papers, not on assumptions.
4.  **Structure the Output:**
    *   Begin with a brief overview of the papers being compared.
    *   Present the points of disagreement in a clear, structured format (e.g., using bullet points or numbered lists).
    *   For each disagreement, clearly state:
        *   The specific point of contention.
        *   The stance of each involved paper (e.g., `[Paper A]` argues X, while `[Paper B]` argues Y).
        *   A brief quote or paraphrase from each paper supporting the identified disagreement.
    *   Conclude with a summary of the overall landscape of agreement and disagreement.

# OUTPUT FORMAT

Present your findings in a structured report. Use Markdown for formatting.

**Report Title:** Comparative Analysis of LLM Capabilities in Literature Review: Points of Disagreement

**Introduction:**
Briefly list the five papers being analyzed (e.g., Paper 1: "Title A", Paper 2: "Title B", etc.).

**Key Areas of Disagreement:**

*   **Disagreement 1: [Specific area of contention, e.g., LLM Accuracy in Reference Generation]**
    *   **[Paper X]:** [Summary of X's stance and supporting quote/paraphrase]
    *   **[Paper Y]:** [Summary of Y's stance and supporting quote/paraphrase]
    *   **[Paper Z]:** [Summary of Z's stance and supporting quote/paraphrase]
    *   *Note on Disagreement:* [Brief analysis of why they disagree or the implication of their differing views]

*   **Disagreement 2: [Another specific area of contention, e.g., Effectiveness of Mutual Constraint Prompting]**
    *   **[Paper A]:** [Summary of A's stance and supporting quote/paraphrase]
    *   **[Paper B]:** [Summary of B's stance and supporting quote/paraphrase]
    *   *Note on Disagreement:* [...]

*(Continue for all identified significant disagreements)*

**Conclusion:**
Summarize the overarching themes of disagreement and potential implications for future research or application.

**Constraint Checklist:**
- [ ] All five papers are analyzed.
- [ ] Points of disagreement are clearly identified.
- [ ] Each disagreement is supported by evidence (quotes/paraphrases) from the respective papers.
- [ ] Papers involved in each disagreement are explicitly named.
- [ ] Output is structured as a report with clear sections.
- [ ] No external information is used; analysis is solely based on the provided papers.