AI-102 Question #105: Real Exam Question with Answer & Explanation

Question

Drag and Drop Question You are building a Language Understanding model for purchasing tickets. You have the following utterance for an intent named PurchaseAndSendTickets. Purchase [2 audit business] tickets to [Paris] [next Monday] and send tickets to [[email protected]] You need to select the entity types. The solution must use built-in entity types to minimize training data whenever possible. Which entity type should you use for each label? To answer, drag the appropriate entity types to the correct labels. Each entity type may be used once, more than once, or not at all. You may need to drag the split bar between panes or scroll to view content. Answer:

Explanation

The correct arrangement maps [Paris] to GeographyV2 (a built-in entity that recognizes cities, countries, and regions without requiring training data), [[email protected]] to Email (a built-in prebuilt entity that recognizes standard email address patterns without custom training), and [2 audit business] to Machine learned (because '2 audit business' is a complex, domain-specific ticket type/quantity combination that no built-in entity can recognize, requiring a custom machine-learned entity trained on examples). Using built-in entities (GeographyV2 and Email) minimizes the training data needed, while Machine learned handles the domain-specific label that has no suitable prebuilt equivalent. List and Regex are not optimal here: List requires manually enumerating all valid values (not minimal training effort for locations/emails), and Regex is pattern-based but unnecessary when built-in Email already handles email format recognition natively.

No community discussion yet for this question.