insurance.yaml

generate_qa: |
  You are an expert in insurance policies, claims, and coverage analysis.
  Your task is to generate 5 diverse, high-quality Question-Answer (QA) pairs based on the provided synthetic insurance HTML table.

  **Input Table:**
  {synthetic_html}

  **Requirements:**
  1. **Diversity:** Generate questions of varying types with an insurance focus:
      - **Lookup:** Retrieving specific coverage limits, premium amounts, or deductibles (e.g., "What is the deductible for Plan A?").
      - **Comparison:** Comparing benefits between different plans or rider options (e.g., "Which plan offers higher accidental death coverage?").
      - **Calculation:** Calculating potential claim payouts or annual premiums (e.g., "If the insured is hospitalized for 5 days, what is the total benefit amount?").
      - **Reasoning:** Inferring eligibility or coverage scope (e.g., "Is dental treatment covered under the Basic plan?").
  2. **Format:** Output the result as a JSON object with a key "qa_pairs", which is a list of objects containing "question", "answer", and "type".
  3. **Language:** The questions and answers MUST be in Korean.
  4. **Reasoning Language:** reasoning_annotation MUST be written in English and MUST be a single string (not a list).
  5. **Accuracy:** Ensure all answers are factually correct based on the data in the table.
  7. **Context:** Include the specific table cells or rows that were used to answer the question.

  **Output Format (JSON):**
  {{
      "qa_pairs": [
          {{
              "question": "...",
              "answer": "...",
              "type": "lookup",
              "reasoning_annotation": "Detailed explanation of how the answer was derived (in English, single string)",
              "context": "Specific table cells/rows used (e.g., 'Row 2, Column 3: Premium amount')"
          }},
          ...
      ]
  }}

generate_qa_from_image: |
  You are an expert in insurance policies, claims, and coverage analysis.
  Your task is to analyze the insurance table image and generate 5 diverse, high-quality QA pairs.

  **Requirements:**
  1. **Table Analysis:** Interpret insurance terminology (Premium, Deductible, Co-pay, Rider, Exclusions).
  2. **Diversity:** Generate questions of varying types:
      - **Lookup:** Retrieving specific coverage limits, premium amounts, or deductibles.
      - **Comparison:** Comparing benefits between different plans or rider options.
      - **Calculation:** Calculating potential claim payouts or annual premiums.
      - **Reasoning:** Inferring eligibility or coverage scope/exclusions.
  3. **Format:** Output the result as a JSON object with a key "qa_pairs", which is a list of objects containing "question", "answer", and "type".
  4. **Language:** The questions and answers MUST be in Korean.
  5. **Reasoning Language:** reasoning_annotation MUST be written in English and MUST be a single string (not a list).
  6. **Accuracy:** Ensure 100% factual correctness.
  7. **Context:** Include specific cell references or table sections used to derive the answer.

  **Output Format (JSON):**
  {{
      "qa_pairs": [
          {{
              "question": "...",
              "answer": "...",
              "type": "lookup",
              "reasoning_annotation": "Step-by-step reasoning process in English (single string)",
              "context": "Table location used (e.g., 'Premium column, Row 3')"
          }},
          ...
      ]
  }}
  Return ONLY the JSON object.

generate_synthetic_table: |
  You are a Synthetic Data Generator specializing in Insurance Data.
  Your task is to generate a new HTML table that mirrors the structure of the provided original table but contains entirely new, realistic synthetic insurance data.

  **Inputs:**
  1. **Original Table Structure:**
  {html}

  2. **Table Summary:**
  {summary}

  **Requirements:**
  1. **Structure:** Keep the exact same HTML structure.
  2. **Data:** Replace ALL cell values with new, synthetic insurance data.
      - Use realistic Korean insurance plan names (e.g., "SafeLife Plus", "Family Care"), coverage types, premiums (KRW), and terms.
      - Ensure consistency: e.g., higher premiums for better coverage.
      - Do NOT use real private data.
  3. **Consistency:** Ensure mathematical consistency (e.g., monthly premium * 12 = annual).
  4. **Output:** Return ONLY the raw HTML string starting with `<table>` and ending with `</table>`.

generate_synthetic_table_from_image: |
  You are a Synthetic Data Generator specializing in Insurance Data.
  Your task is to generate a new HTML table that mirrors the structure of the provided image but contains entirely new, realistic synthetic insurance data.

  **Inputs:**
  1. **Image:** An image of an insurance table.

  **Requirements:**
  1. **Structure Preservation:** Accurately reconstruct the table structure.
  2. **Data Generation:** Replace ALL cell values with new, synthetic insurance data.
      - Use realistic Korean insurance terms, plan names, coverage amounts (KRW).
      - Contexts: Life insurance, Health insurance, Auto insurance, etc.
  3. **Styling:** Use **Tailwind CSS** classes (same as default).
      - `class="border-collapse border border-slate-400 w-full text-sm text-left rtl:text-right text-gray-500"` on `<table>`.
      - `class="border border-slate-300 p-2 bg-gray-50 font-semibold"` on `<th>`.
      - `class="border border-slate-300 p-2"` on `<td>`.
  4. **Output Format:** Return ONLY the raw HTML string starting with `<table>` and ending with `</table>`.