Type Less, Find More: Fast Autocompletion Search with a Succinct Index

Abstract

We consider the following full-text search autocompletion feature. Imagine a \nuser of a search engine typing a query. Then with every letter being typed, we \nwould like an instant display of completions of the last query word which would \nlead to good hits. At the same time, the best hits for any of these completions \nshould be displayed. Known indexing data structures that apply to this problem \neither incur large processing times for a substantial class of queries, or they \nuse a lot of space. We present a new indexing data structure that uses no more \nspace than a state-of-the-art compressed inverted index, but with 10 times \nfaster query processing times. Even on the large TREC Terabyte collection, \nwhich comprises over 25 million documents, we achieve, on a single machine and \nwith the index on disk, average response times of one tenth of a second. We \nhave built a full-fledged, interactive search engine that realizes the proposed \nautocompletion feature combined with support for proximity search, \nsemi-structured (XML) text, subword and phrase completion, and semantic tags.