Synopses of Movie Narratives (SYMON)

SYMON是一个大规模的视频语言数据集，专注于电影叙事的故事理解。该数据集由南洋理工大学和弗吉尼亚大学的研究团队创建，包含5193个流行电影和电视剧的视频摘要，总时长达到869小时。SYMON数据集的特点是覆盖了多模态故事事件，并包含了丰富的角色心理状态描述。数据集的内容代表了自然主义多模态故事讲述技巧，旨在为多模态故事理解提供基础，并解决现有模型面临的跨领域语义差距问题。SYMON数据集的应用领域包括视频文本检索和零样本对齐，旨在推动多模态学习领域的进步。

SYMON is a large-scale video-language dataset focused on story understanding in cinematic narration. It was developed by research teams from Nanyang Technological University and the University of Virginia. The dataset contains 5,193 video summaries of popular films and TV series, with a total duration of 869 hours. SYMON features multimodal story events and incorporates rich descriptions of characters' psychological states. Its content embodies naturalistic multimodal storytelling techniques, and it aims to provide a foundational resource for multimodal story understanding while addressing the cross-domain semantic gap challenges faced by existing models. The SYMON dataset has applications in video-text retrieval and zero-shot alignment, with the objective of advancing the field of multimodal learning.