A Method of Integrating Length Constraints into Encoder-Decoder Transformer for Abstractive Text Summarization  

作　　者：Ngoc-Khuong Nguyen Dac-Nhuong Le Viet-Ha Nguyen Anh-Cuong Le 

机构地区：[1]Faculty of Information Technology,Haiphong University,Haiphong,180000,Vietnam [2]Faculty of Information Technology,VNU University of Engineering and Technology,Hanoi,10000,Vietnam [3]Natural Language Processing and Knowledge Discovery Laboratory,Faculty of Information Technology,Ton Duc Thang University,Ho Chi Minh City,Vietnam

出　　处：《Intelligent Automation & Soft Computing》2023年第10期1-18,共18页智能自动化与软计算（英文）

基　　金：funded by Vietnam National Foundation for Science and Technology Development(NAFOSTED)under Grant Number 102.05-2020.26。

摘　　要：Text summarization aims to generate a concise version of the original text.The longer the summary text is,themore detailed it will be fromthe original text,and this depends on the intended use.Therefore,the problem of generating summary texts with desired lengths is a vital task to put the research into practice.To solve this problem,in this paper,we propose a new method to integrate the desired length of the summarized text into the encoder-decoder model for the abstractive text summarization problem.This length parameter is integrated into the encoding phase at each self-attention step and the decoding process by preserving the remaining length for calculating headattention in the generation process and using it as length embeddings added to theword embeddings.We conducted experiments for the proposed model on the two data sets,Cable News Network(CNN)Daily and NEWSROOM,with different desired output lengths.The obtained results show the proposed model’s effectiveness compared with related studies.

关 键 词：Length controllable abstractive text summarization length embedding 

分 类 号：TP39[自动化与计算机技术—计算机应用技术]