CCA-F: Claude Certified Architect — Foundations

By Anthropic · Question Mode

QuestionQ112

Prompt Engineering & Structured Output

You are developing a structured data-extraction system with Claude. The system extracts information from unstructured documents, validates its output with JavaScript Object Notation (JSON) schemas, and preserves high accuracy. It must gracefully handle edge cases and integrate with downstream systems.

Your extraction system processes e-commerce product descriptions to pull specifications, such as dimensions, weight, and materials, into JSON. Although the schema is well defined, the model extracts the "materials" field inconsistently—sometimes returning "cotton blend", at other times "Cotton/Polyester mix", and occasionally leaving out the field when material information is clearly present in the source.

What is the most effective method to improve extraction consistency?

Explanation

Few-shot examples that pair representative product text with the required canonical material values teach the extraction and normalization convention directly. This addresses both variant phrasing and cases where material information should populate the field. A JSON schema validates structure but does not establish the semantic mapping from source wording to a standardized value. Anthropic’s prompt-engineering guidance notes that Claude pays attention to examples and their details, so examples should consistently reflect the behavior being encouraged.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!