Indexes for document retrieval with relevance

Document Type

Conference Proceeding

Publication Date

1-1-2013

Abstract

Document retrieval is a special type of pattern matching that is closely related to information retrieval and web searching. In this problem, the data consist of a collection of text documents, and given a query pattern P , we are required to report all the documents (not all the occurrences) in which this pattern occurs. In addition, the notion of relevance is commonly applied to rank all the documents that satisfy the query, and only those documents with the highest relevance are returned. Such a concept of relevance has been central in the effectiveness and usability of present day search engines like Google, Bing, Yahoo, or Ask. When relevance is considered, the query has an additional input parameter k, and the task is to report only the k documents with the highest relevance to P, instead of finding all the documents that contains P. For example, one such relevance function could be the frequency of the query pattern in the document. In the information retrieval literature, this task is best achieved by using inverted indexes. However, if the query consists of an arbitrary string - which can be a partial word, multiword phrase, or more generally any sequence of characters - we cannot take advantages of the word boundaries and we need a different approach. This leads to one of the active research topics in string matching and text indexing community in recent years, and various aspects of the problem have been studied, such as space-time tradeoffs, practical solutions, multipattern queries, and I/O-efficiency. In this article, we review some of the initial frameworks for designing such indexes and also summarize the developments in this area. © Springer-Verlag 2013.

Publication Source (Journal or Book title)

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

First Page

351

Last Page

362

This document is currently not available here.

Share

COinS