Ocelot: A System for Summarizing Web Pages

Explore a prototype system called OCELOT designed to automatically generate concise summaries or "gists" of web pages. Learn how this innovative approach tackles the unique challenges of summarizing web content, which often lacks the coherent structure found in traditional text documents like news articles. Discover how OCELOT employs non-extractive summarization techniques, using probabilistic models to select and order words into a concise representation rather than extracting verbatim text spans. Examine the process of training these models using a collection of human-summarized web pages. Gain insights into the complexities of summarizing web content, which frequently consists of a mix of phrases, links, graphics, and formatting commands. This hour-long lecture, presented by Adam Berger from the Center for Language & Speech Processing at Johns Hopkins University, offers a deep dive into cutting-edge text summarization technology specifically tailored for the web.