Utilización de pasajes de tamaño variable para mejorar el proceso de recuperación de información

Abstract

Previous works show that the use of fragments of documents as the basic unit of information to calculate the relevance of a document with regard to a query, improves the results of information retrieval systems. However, it has not been agreed how these fragments of texts should be obtained in order to obtain optimum results. This paper presents an information retrieval system that is based on the definition of passages of variable size. Each passage is formed by a number of sentences that formed the document. The number of sentences by each passage will depend on the position of the query words in each document. The accomplished evaluation allows comparing the performance of this model with a standard information retrieval system, as well as with other passage retrieval systems.