Novel Mining · 小说对话挖掘工作流

What problem does it solve?

This Skill turns raw novel text into structured dialogue data and then derives role-to-role response probabilities and addressee/attitude analyses from the original source, reducing manual annotation and subjective tuning.

Core Features & Use Cases

• Dialogue extraction & speaker annotation: Converts each utterance into JSON with speaker, text, and location metadata (volume/chapter).
• Probability statistics engine: Computes conditional response matrices and silence/interval thresholds to support downstream chat triggering logic.
• Addressee inference & relationship analysis: Infers the intended addressee using rule-based multi-turn reasoning and then quantifies attitude/voice traits to produce relationship matrices.
• Use Case: With the full corpus of the Haruhi Suzumiya novels, generate JSON configuration files and matrices that can power a role-driven multi-character chat engine.

Quick Start

Run the dialogue mining pipeline end-to-end on the full novel text files (volumes 1–13) to produce the annotated dialogue dataset plus probability and relationship matrix outputs.