/abstractive-summarization-kign

Primary LanguagePython

Guiding Generation for Abstractive Text Summarization Based on Key Information Guide Network

Implemented a LSTM-based encoder architecture for learning the representation of keywords extracted using the TextRank algorithm from all the articles in the dataset, and then fed that encoded representation along with the news article to be summarized to an only attention-based pointer-generator network to produce a summary as given in [1]. Further, to guide the network to produce better summaries, the paper implements a prediction-guide mechanism based on [2] which involves using a value estimation network to choose among the best k summaries generated by the beam search decoding process. Successfully reproduced the results of the paper on the CNN/DailyMail news dataset [3].

Implementation based on the paper [4]. Work done as part of the MIDAS Research Lab under the supervision of Dr. Rajiv Ratn Shah.

References

  1. C. Li, W. Xu, S. Li, and S. Gao, “Guiding generation for abstractive text summarization based on key information guide network,” in Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers). New Orleans, Louisiana: Association for Computational Linguistics, Jun. 2018, pp. 55–60. Available here.
  2. D. He, H. Lu, Y. Xia, T. Qin, L. Wang, and T.-Y. Liu, “Decoding with value networks for neural machine translation,” in Proceedings of the 31st International Conference on Neural Information Processing Systems, ser. NIPS’17. USA: Curran Associates Inc., 2017, pp. 177–186. Available here.
  3. K. M. Hermann, T. Kočiský, E. Grefenstette, L. Espeholt, W. Kay, M. Suleyman, and P. Blunsom, “Teaching machines to read and comprehend,” in Proceedings of the 28th International Conference on Neural Information Processing Systems - Volume 1, ser. NIPS’15. Cambridge, MA, USA: MIT Press, 2015, pp. 1693–1701. Available here.
  4. Li, Chenliang, et al. "Guiding generation for abstractive text summarization based on key information guide network." Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers). 2018. Available here.