Karcı summarization: A simple and effective approach for automatic text summarization using Karcı entropy

作者:

Highlights:

• A new, simple and effective approach for automatic summarization of text document using Karcı entropy.

• In the proposed summarization approach, irregularities in the documents is eliminated using a software tool that we call KUSH and the software tool has been developed for text summarization with its unique algorithm.

• Karcı entropy was not used before in any of the previous text summarization studies.

• Experimental results show that the proposed summarizer outperforms all state-of-the-art comparators in terms of the 200-word abstracts in metrics ROUGE-1, ROUGE-2, ROUGE-L, ROUGE-W-1–2. In addition to, proposed summarizer outperforms nearest competitive summarizers 6.4% of ROUGE-1 recall on DUC-2002 dataset. These results show that Karcı summarization is a promising technique and it is expected to attract interest of researchers.

摘要

•A new, simple and effective approach for automatic summarization of text document using Karcı entropy.•In the proposed summarization approach, irregularities in the documents is eliminated using a software tool that we call KUSH and the software tool has been developed for text summarization with its unique algorithm.•Karcı entropy was not used before in any of the previous text summarization studies.•Experimental results show that the proposed summarizer outperforms all state-of-the-art comparators in terms of the 200-word abstracts in metrics ROUGE-1, ROUGE-2, ROUGE-L, ROUGE-W-1–2. In addition to, proposed summarizer outperforms nearest competitive summarizers 6.4% of ROUGE-1 recall on DUC-2002 dataset. These results show that Karcı summarization is a promising technique and it is expected to attract interest of researchers.

论文关键词:Karcı entropy,Entropy,Graph entropy,Graph-based summarization,Generic document summarization;Text mining

论文评审过程:Received 9 July 2019, Revised 22 November 2019, Accepted 18 December 2019, Available online 26 December 2019, Version of Record 26 December 2019.

论文官网地址:https://doi.org/10.1016/j.ipm.2019.102187