Data Cleaning Instruction Prompt AI Prompt

Design a prompt that instructs an LLM to clean and standardize a specific type of messy data field.

Field type: {{field_type}} (e.g. company names, phone numbers, addresses, product descriptions, job titles)
Source data sample: {{data_sample}}

1. The challenge with LLM data cleaning:
   - LLMs are inconsistent without explicit rules — the same model may normalize 'IBM Corp.' differently on two calls
   - The prompt must eliminate ambiguity by providing exhaustive rules and examples

2. Prompt structure for data cleaning:

   a. Task definition (1 sentence): 'Normalize the following {{field_type}} to a standard format.'

   b. Normalization rules (numbered list, in order of priority):
   - Rule 1: [most important normalization, e.g. 'Convert to Title Case']
   - Rule 2: [second rule, e.g. 'Remove legal suffixes: LLC, Inc., Corp., Ltd.']
   - Rule 3: [third rule, e.g. 'Expand common abbreviations: St. → Street, Ave. → Avenue']
   - Continue until all cases are covered

   c. Conflict resolution: 'If two rules conflict, apply the earlier rule.'

   d. Uncertainty handling: 'If you are not confident in the correct normalization, return the input unchanged and append a [?] flag.'

   e. Output format: 'Return ONLY the normalized value. No explanation.'

3. Few-shot examples (critical for consistency):
   - Include 6–10 input → output pairs covering the most common messy patterns
   - Include at least 2 edge cases (very short, very long, non-standard characters)
   - Include 1 example where the model should return the value unchanged with [?]

4. Batch processing version:
   - Extend the prompt to clean a list of 20 values in one call
   - Output as a JSON array preserving input order
   - Include an index field so outputs can be joined back to inputs

Return: single-record cleaning prompt, batch cleaning prompt, test set of 20 messy values, and expected normalized outputs.