IDS 200 Lecture Notes - Lecture 19: Natural-Language Processing, Knowledge Graph, Trade Secret
12 Feb 2020
School
Department
Course
Professor
Document Summary
All the pages google has visited are stored in an internet-scale database system. Page metadata, like creation & revision dates. Document servers; sorting details are an important trade secret. Suppose you have a library full of books and you need to find information on a topic. For every search term, the index holds a list of pages containing the term. With multiple search terms, a relatively small set of webpages will hold all of them. Still, searches for common terms can easily return several million matches . Google is shifting toward a purer semantic approach to providing search results. Determine meaning of search term groups and match to. Content is understood in a combination of term associations and object structures. When a search arrives, google references the terms in its index to find matching pages. The algorithm is a closely-guarded secret but we can talk about some general principles:
