A Decoding Algorithm for Length-Control Summarization Based on Directed Acyclic Transformers

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Huang, Chenyang, Zhou, Hao, Jen, Cameron, Zheng, Kangjie, Zaïane, Osmar R., Mou, Lili
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866913682879938560
author Huang, Chenyang
Zhou, Hao
Jen, Cameron
Zheng, Kangjie
Zaïane, Osmar R.
Mou, Lili
author_facet Huang, Chenyang
Zhou, Hao
Jen, Cameron
Zheng, Kangjie
Zaïane, Osmar R.
Mou, Lili
contents Length-control summarization aims to condense long texts into a short one within a certain length limit. Previous approaches often use autoregressive (AR) models and treat the length requirement as a soft constraint, which may not always be satisfied. In this study, we propose a novel length-control decoding algorithm based on the Directed Acyclic Transformer (DAT). Our approach allows for multiple plausible sequence fragments and predicts a \emph{path} to connect them. In addition, we propose a Sequence Maximum a Posteriori (SeqMAP) decoding algorithm that marginalizes different possible paths and finds the most probable summary satisfying the length budget. Our algorithm is based on beam search, which further facilitates a reranker for performance improvement. Experimental results on the Gigaword and DUC2004 datasets demonstrate our state-of-the-art performance for length-control summarization.
format Preprint
id arxiv_https___arxiv_org_abs_2502_04535
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle A Decoding Algorithm for Length-Control Summarization Based on Directed Acyclic Transformers
Huang, Chenyang
Zhou, Hao
Jen, Cameron
Zheng, Kangjie
Zaïane, Osmar R.
Mou, Lili
Computation and Language
Length-control summarization aims to condense long texts into a short one within a certain length limit. Previous approaches often use autoregressive (AR) models and treat the length requirement as a soft constraint, which may not always be satisfied. In this study, we propose a novel length-control decoding algorithm based on the Directed Acyclic Transformer (DAT). Our approach allows for multiple plausible sequence fragments and predicts a \emph{path} to connect them. In addition, we propose a Sequence Maximum a Posteriori (SeqMAP) decoding algorithm that marginalizes different possible paths and finds the most probable summary satisfying the length budget. Our algorithm is based on beam search, which further facilitates a reranker for performance improvement. Experimental results on the Gigaword and DUC2004 datasets demonstrate our state-of-the-art performance for length-control summarization.
title A Decoding Algorithm for Length-Control Summarization Based on Directed Acyclic Transformers
topic Computation and Language
url https://arxiv.org/abs/2502.04535