Expand description
Structural Emotion Extraction — text → PAD state via structural signals.
Replaces word-list-only valence detection. Instead of matching individual words (“happy” → positive), this module detects what happened in the text:
- Events: commits, builds, errors, fixes, pushes → structural emotions
- Speech acts: questions, imperatives, contemplation → arousal/dominance
- Self-correction: “nem emlékszem”, “igazad van”, “túlkapás” → positive structural signal (learning), not negative
- Context type: code, personal, philosophical, technical → baseline PAD
- Conversational arc: relative arousal change from previous state
Output: PadState + claim text for EmotionalEpisode creation. The episode gets full epistemic backing via the existing system.
Structs§
- Extraction
Result - The result of structural emotion extraction from a text.
Enums§
- Context
Type - Speech
Act - Structural
Event - A structural event detected in text.
Functions§
- extract_
emotion - Extract emotional state from text using structural signals.