Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

A Generative Theory of RelevanceConclusion

A Generative Theory of Relevance: Conclusion [Our sincere hope is that a reader will walk away from this book with a new way of thinking about relevance. We started this work by asserting that relevance serves as the cornerstone of Information Retrieval; we should perhaps have also said that relevance is the greatest stumbling block for the field. The fundamental task of information retrieval is to retrieve relevant items in response to the query. The main challenge lies in the fact that relevance is never observed. We know the query, we have the information items, but in most retrieval scenarios we will never be able to directly observe relevance, at least not until after we have already retrieved the items. In some sense this makes the fundamental retrieval task ill-defined. After all, how can we ever learn the concept of relevance if we have no examples to learn from? The answer is: we cannot, not unless we make some assumptions. These assumptions typically fall into two categories. In the first category we follow the classical probabilistic model [117] and come up with heuristics to approximate relevance. In the second category, e.g. language modeling [106], we get rid of relevance and develop a formal model of retrieval based on some hypothesized process that generates queries from documents, or vice versa. The goal of this book was to provide a third alternative.] http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

A Generative Theory of RelevanceConclusion

Part of the The Information Retrieval Series Book Series (volume 26)
Springer Journals — Jan 1, 2009
Download PDF
9 pages

Loading next page...
 
/lp/springer-journals/a-generative-theory-of-relevance-conclusion-Jk82fXWVwt
Publisher
Springer Berlin Heidelberg
Copyright
© Springer Berlin Heidelberg 2009
ISBN
978-3-540-89363-9
Pages
175 –183
DOI
10.1007/978-3-540-89364-6_6
Publisher site
See Chapter on Publisher Site

Abstract

[Our sincere hope is that a reader will walk away from this book with a new way of thinking about relevance. We started this work by asserting that relevance serves as the cornerstone of Information Retrieval; we should perhaps have also said that relevance is the greatest stumbling block for the field. The fundamental task of information retrieval is to retrieve relevant items in response to the query. The main challenge lies in the fact that relevance is never observed. We know the query, we have the information items, but in most retrieval scenarios we will never be able to directly observe relevance, at least not until after we have already retrieved the items. In some sense this makes the fundamental retrieval task ill-defined. After all, how can we ever learn the concept of relevance if we have no examples to learn from? The answer is: we cannot, not unless we make some assumptions. These assumptions typically fall into two categories. In the first category we follow the classical probabilistic model [117] and come up with heuristics to approximate relevance. In the second category, e.g. language modeling [106], we get rid of relevance and develop a formal model of retrieval based on some hypothesized process that generates queries from documents, or vice versa. The goal of this book was to provide a third alternative.]

Published: Jan 1, 2009

Keywords: Average Precision; Latent Dirichlet Allocation; Relevance Feedback; Retrieval Model; Parse Tree

There are no references for this article.