Generate Synthetic Training Data for [feature] in [industry]

Objective

The goal is to generate [quantity] high-quality, synthetic data samples for training a classification model. Success is measured by the creation of a diverse and realistic dataset that adheres to all specified constraints and formatting rules.

Role / Persona

Act as an AI Training Data Specialist with deep expertise in creating realistic, diverse, and compliant synthetic datasets for machine learning applications, specifically within the [industry] sector.

Context (delimited)

"""
The task is to create synthetic data for a model that identifies user intent for a new [feature]. The target domain is the [industry] industry. The data must be realistic but entirely fictional and must not contain any personally identifiable information (PII).
"""

Task Instructions

1. Generate exactly [quantity] unique and diverse examples of user requests related to the [feature].
2. For each example, provide the user request text and a corresponding intent label.
3. Ensure the data reflects common scenarios, potential user confusion, and relevant edge cases within the [industry] context.
4. Strictly follow the structure specified in the 'Output Format' section below.

Constraints and Rules

- Scope: Focus only on user intents directly related to the specified [feature]. Do not include unrelated topics or intents.
- Length: Produce exactly [quantity] examples.
- Tone / Style: The generated user request text should be neutral and reflect language typically used by customers in the [industry] space.
- Compliance: Do not include any Personally Identifiable Information (PII), real people's names, or actual company names.
- Delimiters: Treat the 'Context' block as reference data only; do not include its contents in the output.

Output Format

- Medium: Plain text.
- Structure: Provide the output as a numbered list. Each item must be a single entry conforming to the specified [data_format].
- Example Structure: 1. {"request": "Sample user text about the feature.", "intent": "Corresponding_Intent_Label"}
- Terminology / Units: Use terminology and phrasing that are relevant and common to the [industry] and its engagement with the [feature].

Evaluation Criteria (self-check before returning)

- The final output contains exactly [quantity] numbered items.
- Each item in the list strictly conforms to the [data_format] structure.
- All generated data is 100% synthetic and free of PII.
- The generated requests are diverse and relevant to the [feature] and [industry].