A Divide-and-Conquer Approach to the Summarization of Long Documents

We present a novel divide-and-conquer method for the neural summarization of long documents. Our method exploits the discourse structure of the document and uses sentence similarity to split the problem into an ensemble of smaller summarization problems... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Text Summarization arXiv DANCER PEGASUS ROUGE-1 45.01 # 2
ROUGE-2 17.6 # 2
ROUGE-L 40.56 # 2
Text Summarization arXiv DANCER RUM ROUGE-1 42.7 # 4
ROUGE-2 16.54 # 3
ROUGE-L 38.44 # 3
Text Summarization arXiv DANCER LSTM ROUGE-1 41.87 # 7
ROUGE-2 15.92 # 4
ROUGE-L 37.61 # 4
Text Summarization Pubmed DANCER LSTM ROUGE-1 44.09 # 5
ROUGE-2 17.69 # 3
ROUGE-L 40.27 # 3
Text Summarization Pubmed DANCER PEGASUS ROUGE-1 46.34 # 1
ROUGE-2 19.97 # 2
ROUGE-L 42.42 # 1
Text Summarization Pubmed DANCER RUM ROUGE-1 43.98 # 6
ROUGE-2 17.65 # 4
ROUGE-L 40.25 # 4

Methods used in the Paper