Significant Sentence Extraction by Euclidean Distance Based on Singular Value Decomposition

Abstract

Abstract. This paper describes an automatic summarization approach that constructs a summary by extracting the significant sentences. The approach takes advantage of the cooccurrence relationships between terms only in the document. The techniques used are principal component analysis (PCA) to extract the significant terms and singular value decompostion (SVD) to find out the significant sentences. The PCA can quantify both the term frequency and term-term relationship in the document by the eigenvalue-eigenvector pairs. And the sentence-term matrix can be decomposed into the proper dimensional sentence-concentrated and term-concentrated marices which are used for the Euclidean distances between the sentence and term vectors and also removed the noise of variability in term usage by the SVD. Experimental results on Korean newspaper articles show that the proposed method is to be preferred over random selection of sentences or only PCA when summarization is the goal.

SemPCA-Summarizer: Exploiting Semantic Principal Component Analysis for Automatic Summary Generation

“…The significant sentences of the text were selected using those thematic words under the hypothesis that those sentences would represent the main topic of a document. An improved approach derived from this research work was later developed in [15], in which PCA was used to identify relevant terms and then, Singular Value Decomposition (SVD) was employed for extracting the significant sentences associated to such terms (the higher number of terms a sentence had, the more important it was considered) and build the final summary. The experiments performed over a set of Korean newspaper articles showed that the method that used PCA and SVD jointly achieved the best performance for F-measure (0.436), compared to the method in which only PCA was employed (0.416).…”

The University of Alicante at MultiLing 2015: approach, results and further insights

“…As a method for developing summarization systems, PCA provides a way to determine the most relevant key terms of a document. It has been often employed in conjunction with other data mining techniques, such as Semantic Vector Space model (Vikas et al, 2008) or Singular Value Decomposition (Lee et al, 2005), using term-based frequency methods. Our main difference with respect to other summarization PCA-based approaches is the incorporation of lexical-semantic knowledge into the PCA technique, since it is necessary to go beyond the terms, and determine the meaningful sentences.…”

Multidocument Summary Generation

“…Document summarization, which aims to create a short version of the original documents, is indispensable yet presents a challenge. Summarization systems can be divided into two categories: (1) single document summarizers [Kupiec et al 1995;Lin and Hovy 1997;Hovy and Marcu 1998;Brunn et al 2001], and (2) multidocument summarizers [Barzilay et al 1999;Chen and Huang 1999;Radev et al 2001;Chen et al 2003a;Lin and Hovy 2002;Lapata 2003;Filatova and Hatzivassiloglou 2004b;Okazaki et al 2004;Sakurai and Utsumi 2004;Lee et al 2005].…”