With the dramatic growth of content on the Web, but also in private
document collections such as intranets and archives, efficieny has become
a major issue in information retrieval applications. Users expect
high-quality results for searches within interactive answer time,
usually below a second for queries on billions of documents.
The talk introduces a number of state-of-the-art algorithms for efficient
query processing in large-scale IR systems. It also covers recent
developments in approximate algorithms that further improve retrieval
efficiency while still providing reasonable result quality.