A Condense-then-Select Strategy for Text Summarization

Hou Pong Chan; Irwin King

RECORD

A Condense-then-Select Strategy for Text Summarization

Hou Pong Chan, Irwin King

Computation and Language · Submitted 19 Jun 2021

abstract compiled · 2026-09-23 · no language-model review

The authors' main claim Experiment results on the CNN/DailyMail, DUC-2002, and Pubmed datasets demonstrate that our framework outperforms the select-then-compress framework and other strong baselines.

how it is stated · asserted 1.00 read the record →

Comment (0) Read on arXiv ↗

Cite ▾

Claim paper

The compiled record

The paper's own sentences, arranged. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.

Open the compiled record

Compiled by classifier-1.13.0 via TypeSafe API, question bank abstract-p1-20260923, 2026-09-23: 1 request, 7,355 input tokens.

What the abstract itself states, as probabilities: claims first 0.13 · states limitation 0.23 · human participants 0.03 · controlled comparison 0.89 · randomized assignment 0.01 · states quantitative result 0.06 · compares with prior results 0.98 · mentions code or data availability 0.02.

The abstract, sentence by sentence

arXiv listing abstract v1. Results and the main claim are underlined; show every role.

Select-then-compress is a popular hybrid, framework for text summarization due to its high efficiency. This framework first selects salient sentences and then independently condenses each of the selected sentences into a concise version. However, compressing sentences separately ignores the context information of the document, and is therefore prone to delete salient information. To address this limitation, we propose a novel condense-then-select framework for text summarization. Our framework first concurrently condenses each document sentence. Original document sentences and their compressed versions then become the candidates for extraction. Finally, an extractor utilizes the context information of the document to select candidates and assembles them into a summary. If salient information is deleted during condensing, the extractor can select an original sentence to retain the information. Thus, our framework helps to avoid the loss of salient information, while preserving the high efficiency of sentence-level compression. Experiment results on the CNN/DailyMail, DUC-2002, and Pubmed datasets demonstrate that our framework outperforms the select-then-compress framework and other strong baselines.

background contribution method result interpretation main claim

What the classifier reports for each sentence. The number is the probability it gave the label.
#	sentence	role	the authors' own work	states the main conclusion	strength
s0	Select-then-compress is a popular hybrid, framework for text summarization due to its high efficiency.	background 1.00	0.03	0.03	asserted 1.00
s1	This framework first selects salient sentences and then independently condenses each of the selected sentences into a concise version.	background 0.96	0.04	0.03	asserted 1.00
s2	However, compressing sentences separately ignores the context information of the document, and is therefore prone to delete salient information.	gap 0.49	0.08	0.04	asserted 0.86
s3	To address this limitation, we propose a novel condense-then-select framework for text summarization.	contribution 1.00	0.67	0.08	asserted 0.93
s4	Our framework first concurrently condenses each document sentence.	method 0.98	0.71	0.03	asserted 0.98
s5	Original document sentences and their compressed versions then become the candidates for extraction.	method 1.00	0.55	0.03	asserted 1.00
s6	Finally, an extractor utilizes the context information of the document to select candidates and assembles them into a summary.	method 1.00	0.65	0.07	asserted 0.99
s7	If salient information is deleted during condensing, the extractor can select an original sentence to retain the information.	method 0.87	0.33	0.12	asserted 0.52
s8	Thus, our framework helps to avoid the loss of salient information, while preserving the high efficiency of sentence-level compression.	interpretation 0.64	0.53	0.41	asserted 0.82
s9	Experiment results on the CNN/DailyMail, DUC-2002, and Pubmed datasets demonstrate that our framework outperforms the select-then-compress framework and other strong baselines.	result 1.00	0.95	0.68	asserted 1.00

What kind of paper the classifier reports this is: method or tool 1.00 · evidence from observation or experiment 0.98.

What was examined, and what was not

sources read

arXiv listing abstract v1

not examined

full text
figures and tables
references and citing sentences
relations to other papers

levels run

P abstract

not run

S sections
T body sentences
C citations
R cross-paper relations

Every quoted sentence on this record is the authors'. Every label is from Pith's closed vocabulary, with the probability shown beside it. Every number in the measurements table is the authors', checked by code. Record as JSON

The paper

Original abstract

Select-then-compress is a popular hybrid, framework for text summarization due to its high efficiency. This framework first selects salient sentences and then independently condenses each of the selected sentences into a concise version. However, compressing sentences separately ignores the context information of the document, and is therefore prone to delete salient information. To address this limitation, we propose a novel condense-then-select framework for text summarization. Our framework first concurrently condenses each document sentence. Original document sentences and their compressed versions then become the candidates for extraction. Finally, an extractor utilizes the context information of the document to select candidates and assembles them into a summary. If salient information is deleted during condensing, the extractor can select an original sentence to retain the information. Thus, our framework helps to avoid the loss of salient information, while preserving the high efficiency of sentence-level compression. Experiment results on the CNN/DailyMail, DUC-2002, and Pubmed datasets demonstrate that our framework outperforms the select-then-compress framework and other strong baselines.

Human review

How Pith checks papers → · Full audit of this paper →

How the literature uses it

What this paper does with the arXiv papers it cites

Each citing sentence is taken from the citing paper's LaTeX source and typed by Pith's classifier. A label appears at probability 0.8 or more (0.95 for support), a contest at 0.7 or more; the quoted words are the citing authors'. How each label was checked · JSON

What it cites · 1 sentence without a confident label

Uses the method or tool of 2 papers

We adopt the diverse beam search algorithm [ref] as the decoding algorithm for an abstractor.

A Simple, Fast Diverse Decoding Algorithm for Neural Generation · Implementation Details · 1.00
We use the Adam optimization algorithm [ref] with an initial learning rate of 5e-4 for ML training and an initial learning rate of 5e-5 for RL training.

Adam: A Method for Stochastic Optimization · Implementation Details · 1.00

Cites as background 3 papers

Sequence labelling models [ref] first learn a relevance score for each document sentence, and then select the document sentences with a score larger than 0.5.

Classify or Select: Neural Architectures for Extractive Document Summarization · Related Work · 0.88
Recent abstractive methods [ref] fine-tune large pre-trained language models on text summarization datasets and achieve state-of-the-art performance.

UniLMv2: Pseudo-Masked Language Models for Unified Language Model Pre-Training · Related Work · 0.96
Abstractive methods rely on the encoder-decoder neural model [ref] to generate a summary.

Neural Machine Translation by Jointly Learning to Align and Translate · Related Work · 0.96

Discussion

Know something about this paper? Leave a comment, or highlight any passage above to comment on it.