Query expansion algorithms based on a new discrete relevance function
No Thumbnail Available
Date
Advisors
Evaluators
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad Industrial de Santander
Abstract
Description
Se ha demostrado que el proceso de expansión de las consultas en el modelo espacio vectorial de representación dedocumentos en un sistema de recuperación de información, es una técnica útil para mejorar la relevancia medidapor la precisión de los resultados entregados a los usuarios. En este artículo se presenta un nuevo algoritmo y unavariación del mismo para realizar expansión de consultas en un sistema de recuperación de información. Estosalgoritmos se basan en una nueva función discreta que define la importancia relativa de un término en una colecciónde documentos. El algoritmo y su variación se evalúan frente a la búsqueda por similitud de cosenos y el algoritmode expansión propuesto por Rocchio, obteniendo excelentes resultados sobre la colección de datos CACM (artículospublicados en la revista Communications of the ACM).
It has been shown that the query expansion process in the vector space model of document’s representation in aretrieval system, it is a useful technique for improving the relevance measured by precision of the results delivered tousers. This paper presents a new algorithm and a variation of itself used to perform query expansion in informationretrieval systems. These algorithms are based on a new discrete function that defines the relative importance of aterm in a document collection. The algorithm and its variation were evaluated against the cosine similarity searchand the query expansion algorithm proposed by Rocchio, with excellent results on data collection CACM (articlespublished in the Communications of the ACM journal).
It has been shown that the query expansion process in the vector space model of document’s representation in aretrieval system, it is a useful technique for improving the relevance measured by precision of the results delivered tousers. This paper presents a new algorithm and a variation of itself used to perform query expansion in informationretrieval systems. These algorithms are based on a new discrete function that defines the relative importance of aterm in a document collection. The algorithm and its variation were evaluated against the cosine similarity searchand the query expansion algorithm proposed by Rocchio, with excellent results on data collection CACM (articlespublished in the Communications of the ACM journal).
Keywords
Query expansion, rocchio, relevant term, IDF, inverse document frequency, Expansión de consulta, rocchio, término relevante, IDF, frecuencia invertida de documento