A customer support ticket classification schema has an enum field `priority` with values `"low"`, `"medium"`, and `"high"`. Support tickets sometimes have conflicting priority signals. What improvement to the schema best handles these ambiguous cases?

Question

Accepted Answer

B. Include `"unclear"` as an additional enum value, giving Claude a valid way to signal ambiguity rather than forcing an incorrect categorization. An `"unclear"` enum value gives Claude a structured, valid way to signal uncertainty for ambiguous cases rather than forcing a potentially incorrect categorization. Production systems can route `"unclear"` tickets to human review for correct classification. Option C breaks structured output parsing. Option D introduces self-reported confidence, which is poorly calibrated in LLMs and is an exam anti-pattern. Option A adds granularity but doesn't handle genuine ambiguity.

A customer support ticket classification schema has an enum field `priority` with values `"low"`, `"medium"`, and `"high"`. Support tickets sometimes have conflicting priority signals. What improvement to the schema best handles these ambiguous cases?

Related Questions

Discussion

A customer support ticket classification schema has an enum field priority with values "low", "medium", and "high". Support tickets sometimes have conflicting priority signals. What improvement to the schema best handles these ambiguous cases?

Related Questions

Discussion

A customer support ticket classification schema has an enum field `priority` with values `"low"`, `"medium"`, and `"high"`. Support tickets sometimes have conflicting priority signals. What improvement to the schema best handles these ambiguous cases?