The authors present an explication interface for event-based emotion analysis that maps input text to a script in the closed vocabulary of Natural Semantic Metalanguage. A fixed decision list of rules transcribed from published semantic definitions computes the label from the explication alone, providing a causal and definitional faithfulness guarantee.
- The parser is fine-tuned to reach 0.33 accuracy and 0.48 selective accuracy on a small held-out set of crowd-sourced event descriptions.
- The system trades insignificant accuracy difference compared to black-box models for a verifiable, inspectable decision basis.
- The authors release EmoExpl-1200 with per-line verification metadata and the full rule set.
This approach allows for auditable first-person event-based emotion analysis by making the per-line entailment interface transparent against the input.