Please use this identifier to cite or link to this item:
Title: Encoding Position Improves Recurrent Neural Text Summarizers
Authors: Karanikolos, Apostolos
Refanidis, Ioannis
Editors: Abbas, Mourad
Freihat, Abed Alhakim
Type: Conference Paper
Subjects: FRASCATI::Natural sciences::Computer and information sciences
Keywords: natural language processing
abstractive text summarization
neural sequence to sequence models
positional embeddings
Issue Date: Sep-2019
Publisher: The Association for Computational Linguistics
First Page: 143
Last Page: 150
Volume Title: Proceedings of the 3rd International Conference on Natural Language and Speech Processing (ICNLSP 2019)
Abstract: Modern text summarizers are big neural networks (recurrent, convolutional, or transformers) trained end-to-end under an encoder-decoder framework. These networks equipped with an attention mechanism, that maintains a memory of their source hidden states, are able to generalize well to long text sequences. In this paper, we explore how the different modules involved in an encoder-decoder structure affect the produced summary quality as measured by ROUGE score in the widely used CNN/Daily Mail and Gigaword summarization datasets. We find that encoding the position of the text tokens before feeding them to a recurrent text summarizer gives a significant, in terms of ROUGE, gain to its performance on the former but not the latter dataset.
Electronic ISBN: 978-1-950737-62-8
Appears in Collections:Department of Applied Informatics

Files in This Item:
File Description SizeFormat 
paper_8.pdfPreprint889,59 kBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.