PerLTQA

PerLTQA数据集由香港中文大学开发，专注于个人长期记忆在问答任务中的分类、检索和合成。该数据集包含8,593个问题，涉及30个角色，涵盖世界知识、个人资料、社交关系、事件和对话等多种记忆类型。数据集的创建旨在探索个性化记忆在社交互动和事件中的应用，特别是在大型语言模型中的应用。通过精细的分类、检索和合成机制，PerLTQA支持对个性化记忆的深入研究和应用，为提升对话系统的记忆处理能力提供了重要资源。The PerLTQA dataset, developed by The Chinese University of Hong Kong, focuses on the classification, retrieval and synthesis of personal long-term memory within the context of question answering tasks. It contains 8,593 questions involving 30 characters, covering a variety of memory types including world knowledge, personal profiles, social relationships, events and conversations. The dataset was developed to explore the applications of personalized memory in social interactions and events, particularly in large language models (LLMs). Through its refined classification, retrieval and synthesis mechanisms, PerLTQA supports in-depth research and applications of personalized memory, providing an important resource for enhancing the memory processing capabilities of dialogue systems.