Sentence Reconstruction in Summary Generation: An Investigation using Automated Alignment.

Accession number;02A0233741
Title;Sentence Reconstruction in Summary Generation: An Investigation using Automated Alignment.
Author; TAKEUCHI KAZUHIRO (Advanced Inst. Sci. and Technol., Nara) MATSUMOTO YUJI (Advanced Inst. Sci. and Technol., Nara)
Journal Title;Joho Shori Gakkai Kenkyu Hokoku
Journal Code:Z0031B
ISSN:0919-6072
VOL.2002;NO.4(NL-147);PAGE.21-28(2002)
Figure&Table&Reference;FIG.4, TBL.7, REF.6
Pub. Country;Japan
Language;Japanese
Abstract;In this paper, we investigate operations in summary generation. In order to align summary expressions with their corresponding original expressions in source text, we propose an automated alignment algorithm based on dependency structure of sentences. Our algorithm detects not only one-to-one sentence alignments, but also one-to-n sentence alignments. We apply the algorithm to human made natural summaries, and analyze the results of the alignments. We, then, find most of the summary expressions are kept their dependency structure in the original sentences and confirm one of the opration called "sentence combination," in which more than two source sentences are used to generate a summary sentence, plays an important role in summary generation. Furthermore, we characterize paraphrases in summary generation. (author abst.)