University of Vienna
Current search engines process thousands of queries per second over a collection of billions of web pages with a sub-second average response time. There are two reasons for this astonishing performance: Massive parallelism and a simple yet efficient data structure, called inverted index.
In this chapter we will describe inverted indices. The parallelism deployed by search engines is quite straightforward: Given a collection of documents and a user query the goal of information retrieval ...