Newspaper Navigator

Newspaper Navigator 的目标是重新构想美国编年史中视觉内容的搜索。该项目包括两个阶段：
通过使用新兴的机器学习技术从编年史的 1630 万份历史报纸页面中提取标题、照片、插图、地图、漫画、卡通和广告，创建 Newspaper Navigator 数据集。除了视觉内容外，数据集还包括从 METS/ALTO OCR 派生的字幕和其他相关文本，以及用于快速相似性查询的图像嵌入。
为 Newspaper Navigator 数据集创建探索性搜索应用程序，以便为美国公众提供浏览编年史美国的新方法。

The goal of Newspaper Navigator is to reimagine the search for visual content within the Chronicling America archive. This project consists of two phases: First, create the Newspaper Navigator dataset by extracting headlines, photographs, illustrations, maps, comics, cartoons, and advertisements from 16.3 million historical newspaper pages of the Chronicling America archive using emerging machine learning techniques. In addition to visual content, the dataset also includes captions and other relevant text derived from METS/ALTO OCR, as well as image embeddings for fast similarity queries. Second, develop an exploratory search application for the Newspaper Navigator dataset to provide the American public with novel ways to explore the Chronicling America archive.

搜集汇总

背景与挑战

背景概述

Newspaper Navigator是一个基于美国历史报纸的数据集，利用机器学习从1630万份页面中提取视觉内容和相关文本，旨在创建探索性搜索应用程序以改进公众浏览体验。