Nowdays Internet search engines such as Google, Yahoo or MSN
Search are ubiquitous. They rely on efficient text retrieval techniques as the WWW consists of bilions of documents already.
The core of every Internet search engine is the index and one of the fundamental index processing techniques are top-k queries. I will present two basic algorithms for evaluating them: No Random Access Algorithm (NRA) and Combined Algorithm (CA).
The aim of the presentation is to introduce a method of queue vector scheduling that accelerates the query processing. The method is based on selectivity estimators and score predictions. It has been experimentally evaluated based on a large Web corpus and tested for efficiency and recall measures.