Abstractive Text Summarization Using Hybrid Technique of Summarization

被引：2

作者：

Liaqat, Muhammad Irfan ^{[1
]}

Hamid, Isma ^{[1
]}

Nawaz, Qamar ^{[2
]}

Shafique, Nida ^{[1
]}

机构：

[1] Natl Text Univ, Dept Comp Sci, Faisalabad, Pakistan

[2] Univ Agr Faisalabad, Dept Comp Sci, Faisalabad, Pakistan

来源：

2022 14TH INTERNATIONAL CONFERENCE ON COMMUNICATION SOFTWARE AND NETWORKS (ICCSN 2022) | 2022年

关键词：

abstractive summarization; BERT2BERT; ParsBERT; Seq-to-Seq;

D O I：

10.1109/ICCSN55126.2022.9817599

中图分类号：

TP31 [计算机软件];

学科分类号：

081202 ; 0835 ;

摘要：

text summarization focuses at compress the document into a shorter form while keeping the connotation intact. The extractive summary can select chunks of sentences that are very related to the document, on the other hand, an abstractive summary can generate a summary based on extracted keywords. This research proposed an abstractive text summarization model, it gets data from source data (e.g., Daily Mail/CNN) or other documents and two summaries of this are generated. one from a philologist and the other by proposed model. The summary generated by the philologist kept as a model to compare with the machine-generated summary. The proposed model increased the accuracy and the readability of the summary.

引用

页码：141 / 144

页数：4

共 10 条

[1]

Ethayarajh K, 2019, Arxiv, DOI arXiv:1909.00512

[2] Recent automatic text summarization techniques: a survey [J].

Gambhir, Mahak ;

Gupta, Vishal .

ARTIFICIAL INTELLIGENCE REVIEW, 2017, 47 (01) :1-66

[3]

Khandelwal U., SAMPLE EFFICIENT TEX

[4]

Lewis M, 2019, Arxiv, DOI [arXiv:1910.13461, 10.48550/arXiv.1910.13461]

[5]

Moratanch N, 2016, PROCEEDINGS OF IEEE INTERNATIONAL CONFERENCE ON CIRCUIT, POWER AND COMPUTING TECHNOLOGIES (ICCPCT 2016)

[6]

Nallapati R, 2016, Arxiv, DOI arXiv:1602.06023

[7]

Su M.-H., INT J NEUROSCI, V28, P2061

[8]

Wang AL, 2019, Arxiv, DOI arXiv:1812.10860

[9] A Topic Information Fusion and Semantic Relevance for Text Summarization [J].

You, Fucheng ;

Zhao, Shuai ;

Chen, Jingjing .

IEEE ACCESS, 2020, 8 :178946-178953

[10]

Zheng Ce, 2020, arXiv

← 1 →