6.3.1.2. Entity Extraction

6.3.1.2. Entity Extraction

Entities are the specific pieces of information within utterances that your application needs to extract. In "Book a flight to Paris tomorrow," the entities are "Paris" (destination) and "tomorrow" (date).

Entity types in CLU:

• Learned entities: ML-based extraction from labeled examples
• List entities: Exact match from predefined values (e.g., product names)
• Prebuilt entities: Ready-to-use extractors (datetime, number, email)
• Regex entities: Pattern-based extraction (order IDs, codes)

When to use which:

• Learned: Flexible extraction when patterns vary
• List: Closed set of known values (cities, products)
• Prebuilt: Standard types (dates, numbers)
• Regex: Structured patterns (ABC-12345 format)