新华日报历史报全量数据集

汇聚超3万站点、1.7亿条数据，沉淀1800余个专题库，覆盖六大模态数据，构建了包括《新华日报》历史报全量数据集、智能媒体资源数据集、深度报道长文档数据集等10个高质量数据集。通过本地化算法自主训练与多模态检索技术，实现“以文搜图/视频”的秒级响应，结合可视化分析工具，全面支撑采编运营决策，显著提升内容生产与传播效能。This dataset aggregates over 30,000 sites, 170 million data entries and has accumulated more than 1,800 specialized repositories, covering six modalities of data. It has developed 10 high-quality datasets including the full historical archive dataset of Xinhua Daily, intelligent media resource dataset, long-form in-depth reporting document dataset and others. Through autonomously trained localization algorithms and multi-modal retrieval technologies, it achieves second-level response for "text-to-image/video search". Combined with visual analysis tools, it fully supports news reporting, editing and operational decision-making, and significantly improves the efficiency of content production and dissemination.

搜集汇总

背景与挑战

背景概述

该数据集是《新华日报》历史报纸的全面集合，包含超3万站点、1.7亿条数据，覆盖1800余个专题库和六大模态数据，并整合了智能媒体资源、深度报道长文档等10个高质量子集。它通过本地化算法和多模态检索技术实现“以文搜图/视频”的秒级响应，结合可视化工具，有效支持采编决策，提升内容生产与传播效率。