Personality-aware Human-centric Multimodal Reasoning Dataset (PHMRD)

PHMRD数据集由南京理工大学计算机科学与工程学院创建，专注于个性感知的人本多模态推理任务。该数据集包含225个角色和12,000个样本，源自六部电视剧。数据集通过PDB2网站获取角色的个性特征，并将其作为个性标注。PHMRD旨在通过整合过去实例的多模态信息和个性因素，预测特定个体在复杂社交互动中的未来行为。该数据集适用于开发和评估个性感知的多模态推理模型，特别是在预测个体行为和心理状态方面。The PHMRD dataset was developed by the School of Computer Science and Engineering, Nanjing University of Science and Technology, focusing on person-centric multimodal reasoning tasks with personality awareness. It contains 225 characters and 12,000 samples derived from six television series. The dataset collects character personality traits from the PDB2 website, which are utilized as personality annotations. The core goal of PHMRD is to predict the future behaviors of specific individuals in complex social interactions by integrating multimodal information and personality factors from previous instances. This dataset is applicable to the development and evaluation of personality-aware multimodal reasoning models, especially for forecasting individual behaviors and psychological states.