Chinese Personalized and Emotional Dialogue

CPED，全称中文个性化和情感对话，是第一个大规模的中文个性化和情感对话数据集。数据集由同理心和个人特征相关的多源知识组成 (涵盖性别、五种人格特征、13种情绪、19种对话行为和10种情景等)。
133,000多模式上下文话语来自40台电视中392位说话者的12,000多次对话显示了3个字符属性 (姓名，性别，年龄) 注释，五种人格特质注释，2种动态情感信息 (情感和情感) 注释和DA注释三个任务: 对话中的人格识别 (PRC)，对话中的情感识别 (ERC) 和个性化和情感对话 (PEC)。

CPED, whose full name is Chinese Personalized and Emotional Dialogue, is the first large-scale Chinese personalized and emotional dialogue dataset. The dataset consists of multi-source knowledge related to empathy and personal characteristics, covering gender, five major personality traits, 13 types of emotions, 19 dialogue acts, 10 scenarios, and more. Over 133,000 multimodal contextual utterances derived from 12,000+ conversations involving 392 speakers across 40 television programs are annotated with three character attributes (name, gender, age), five sets of personality trait annotations, two types of dynamic emotional information (affect and emotion) and dialogue act (DA) annotations, supporting three downstream tasks: Personality Recognition in Conversation (PRC), Emotion Recognition in Conversation (ERC), and Personalized and Emotional Conversation (PEC).

搜集汇总

背景与挑战

背景概述

CPED（中文个性化和情感对话）是首个大规模中文个性化和情感对话数据集，包含133,000多模式上下文话语，来自40台电视中392位说话者的12,000多次对话，并注释了人格特征、情绪、对话行为等多源知识。该数据集设计用于支持三个核心任务：对话中的人格识别、情感识别以及个性化和情感对话生成，适用于自然语言处理领域的研究和应用。