What's Next? Index Structures for Efficient Phrase Querying

Text retrieval systems are used to fetch documents from large text collections, using queries consisting of words and word sequences. A shortcoming of current systems is that word-sequence queries, also known as phrase queries, can be expensive to evaluate, particularly if they include common words. Another limitation is that some forms of querying are not supported; an example is phrase completion, which provides an alternative way of locating information. We propose a new index structure, a nextword index, that addresses both of these problems. We show experimentally that nextword indexes can be used for rapid phrase querying, and show that they allow practical phrase completion.