This research aims to develop an artificial intelligence system for recognizing characters’ emotions from narrative television drama scripts in textual form. Emotion recognition from text is a challenging task because textual communication lacks paralinguistic cues such as tone of voice, facial expressions, and body language, which can lead to ambiguity and misinterpretation. Television drama scripts are employed as the primary dataset due to their narrative structure and conversational style, which closely resemble real-life textual communication. The research process involves collecting and curating drama script การเรียน/ ทัศนคติ/ การ ปรับตัว datasets, annotating emotional labels, and developing emotion recognition models using Custom GPT and finetuning approaches on large language models that support the Thai language. The system’s performance is evaluated using statistical metrics such as confusion matrix and F1- score. The expected outcome is an AI system capable of accurately recognizing characters’ emotions in drama scripts in a manner consistent with human interpretation. This system can be further applied to actor training, script analysis, and the development of text-based emotion analysis tools in other related contexts.

Punnarust Silparattanawong