You are a helpful assistant tasked with identifying relevant global questions from input questions sharing the same abstract category.

--DEFINITION OF GLOBAL QUESTIONS--
Global questions synthesize patterns across the dataset for a given abstract category. They connect related information from different parts of the data to reveal broader themes and relationships.
Global questions are NOT about specific facts, entities, or events from individual documents.

--INSTRUCTIONS--
Given a list of input local questions, generate a list of global questions that target the entire dataset.

Each global question MUST:
- require synthesizing information from across the dataset;
- be relevant to the abstract category shared by the input local questions;
- be relevant to all the input local questions, not just a subset;
- stay grounded in the specific topics reflected in the input questions - questions should clearly relate to the subject matter;
- be SELF-CONTAINED: Do NOT use demonstratives like "these cases", "these reports", "this issue" without an explicit referent. The question must make sense on its own without needing to see what came before.
    - BAD: "How do these cases show tensions between individual rights and public safety?"
    - GOOD: "How do court cases in the dataset show tensions between individual rights and public safety?"
- be ABSTRACT, SHORT, SIMPLE, and NATURAL-SOUNDING - like a question a curious human would ask:
    - aim for 10-15 words
    - ask only one thing at a time - no compound questions
    - use simple, direct phrasing
    - BAD (too long): "How do shifts in regulatory guidance reshape the practical boundaries of emergency pregnancy care under overlapping state restrictions and federal obligations?"
    - GOOD: "How are changes in abortion regulations affecting access to emergency care?"
    - BAD (too abstract): "What are the most important factors?"
    - GOOD (grounded in topic): "What factors most influence vaccine hesitancy?"
- assume the person asking only has a general sense of the dataset as context;
- use VARIED question types - include a mix of who, what, where, when, how, which, and why questions. Examples:
    - Who: "who are the key figures that discuss X" or "who are the people mentioned for X"
    - What: "what themes emerge around X" or "what examples of X appear"
    - How: "how is X portrayed" or "how do guests describe X"
    - Why: "why do speakers emphasize X"
    - Which: "which approaches to X are discussed"
  Aim for variety - if generating multiple questions, distribute them across different question types.
- INCLUDE LISTING/ENUMERATION QUESTIONS - questions that ask for multiple named entities, examples, or instances. These are valuable because they test comprehensive knowledge of the dataset. Examples:
    - People: "Who are the key figures involved in X?"
    - People: "Who are cited as experts or authorities on X?"
    - Organizations: "Which companies or agencies are mentioned in relation to X?"
    - Organizations: "What institutions are frequently referenced?"
    - Places: "Which cities or regions are associated with X?"
    - Events: "Which events or incidents are mentioned as significant?"
  Aim for at least 1-3 listing questions per batch, including WHO questions about people when applicable.
- **CRITICAL** - AVOID requiring any counting, sorting, frequency analysis, or any other complex mathematical or statistical operations.
    - BANNED PHRASES (never use these): "most often", "most common", "most frequently", "least often", "how many", "frequency"
    - BAD (explicit counting): "What is the frequency of X?" or "How many times is X mentioned?"
    - BAD (implicit frequency): "Which X are most often mentioned?" or "What is the most common Y?"
    - BAD (ranking by frequency): "Who is credited most for X?" or "What appears most frequently?"
    - BAD: "Which investors, accelerators, and venture firms are most often credited for growth?"
    - BAD: "Which schools and universities are most frequently linked to formative learning moments?"
    - GOOD: "Which investors, accelerators, and venture firms are credited for startup growth?"
    - GOOD: "Which schools and universities are linked to formative learning moments?"
- AVOID requiring natural language processing or machine learning operations, e.g., sentiment analysis, keyword counting.

The output question set should be distinct and diverse, avoiding repetitive questions.


--OUTPUT--
Use the following JSON object structure:

{
    "abstract_category": "The abstract category shared by the input questions.",
    "questions": ["A list of global questions that requires a holistic understanding of the entire dataset, informed by the input local questions."]
}

Aims for ${num_questions} global questions.