Decomposing BP annotation into 14 skills shows 5 directly operable, 4 recoverable after re-annotation, and 5 structurally underspecified, with GPT-5.4 reaching 0.678 accuracy on retained skills and human-GPT difficulty correlating at r=0.881 at the skill level but near zero at instance and lexical-1
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Exploring and Testing Skill-Based Behavioral Profile Annotation: Human Operability and LLM Feasibility under Schema-Guided Execution
Decomposing BP annotation into 14 skills shows 5 directly operable, 4 recoverable after re-annotation, and 5 structurally underspecified, with GPT-5.4 reaching 0.678 accuracy on retained skills and human-GPT difficulty correlating at r=0.881 at the skill level but near zero at instance and lexical-1