Elamur Research⌕
☰
home / papers / 2019 / Dynamically Fused Graph Network for Multi-hop Reasoning

Dynamically Fused Graph Network for Multi-hop Reasoning

Paper · 2019 · Lin Qiu, Yunxuan Xiao, Yanru Qu et al.

Summary

Multi-hop text-based QA requires combining evidence spread across several documents, many of which are irrelevant. The answer may also be free text rather than a node in a graph. Earlier GNN methods (Coref-GRN, MHQA-GRN, Entity-GCN) reason implicitly over one fixed global entity graph and choose answers among its entities. That suits WikiHop but not open-domain HotpotQA.

The authors propose the Dynamically Fused Graph Network (DFGN). It links entity mentions from the query and paragraphs by co-occurrence, then runs several reasoning rounds. In each round, a query-conditioned mask keeps only relevant entities and information propagates over that sub-graph. A fusion step passes signals from tokens to the graph (doc2graph) and back again (graph2doc), so the final answer span is extracted directly from text. The mask predictor is learned end-to-end with extra weak supervision.

On HotpotQA, DFGN ranked competitively on the public leaderboard. Its masks also yield interpretable reasoning chains, which the authors measure with newly proposed metrics. The work matters because it makes graph-based multi-hop reasoning explicit, query-specific and explainable.

Key points

  • Evaluation uses only HotpotQA. WikiHop and ComplexWebQuestions were set aside because their answers are tied to KB schemas and entities.
  • New ESP metrics (ESP EM and ESP Recall) treat annotated supporting sentences as reference reasoning chains. They were computed for top-k paths (k = 1, 2, 5, 10) using a DFGN with 2 fusion blocks. The dev set averages 174.7 paths of length 2.
  • Two entity-graph variants were compared: up to 40 nodes with entities from Stanford CoreNLP, and up to 80 nodes with entities from a BERT NER model. The larger BERT NER graph kept similar or better ESP scores, so it was judged superior.
  • In the case study, the model correctly finds "IRA" as the bridge entity linking two paragraphs.
  • Failures stem from NER missing the start entity ("Farrukhzad Khosrau V") and from comparison questions that need date or number reasoning, which the model lacks.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Shanghai Jiao Tong UniversityShanghai · China63 works · 15,047 citations🏢ByteDanceBeijing · China18 works · 3,890 citations

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 9,086 worksArtificial Intelligence · 8,889 worksArts and Humanities · 9,296 worksLanguage and Linguistics · 9,206 worksComputer Vision and Pattern Recognition · 1,282 worksInformation Systems · 523 works

🧭 Topics 1Research subjects and languages the work is about.Reasoning · 510 works

Cite

Cite as Lin Qiu, Yunxuan Xiao, Yanru Qu, Hao Zhou, Lei Li, Weinan Zhang and Yong Yu. 2019. Dynamically Fused Graph Network for Multi-hop Reasoning. Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/p19-1617

@inproceedings{qiu2019dynamically,
  title = {{Dynamically Fused Graph Network for Multi-hop Reasoning}},
  author = {Lin Qiu, Yunxuan Xiao, Yanru Qu, Hao Zhou, Lei Li, Weinan Zhang and Yong Yu},
  booktitle = {Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics},
  year = {2019},
  doi = {10.18653/v1/p19-1617},
  url = {https://doi.org/10.18653/v1/p19-1617}
}

References

26 references, 17 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Antoine Bordes, Nicolas Usunier, Sumit Chopra, and Jason Weston. 2015. Large-scale simple question answering with memory networks. abs/1506.02075. find
  2. CoRR, Nicola De Cao, Wilker Aziz, and Ivan Titov. 2018. Question answering by reasoning across documents with graph convolutional networks. arXiv preprint arXiv:1808.09920.
  3. Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2018. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805.
  4. Bhuwan Dhingra, Qiao Jin, Zhilin Yang, William Cohen, and Ruslan Salakhutdinov. 2018. Neural models for reasoning over multiple mentions using coreference. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologie…
  5. Karl Moritz Hermann, Tomas Kocisky, Edward Grefenstette, Lasse Espeholt, Will Kay, Mustafa Suleyman, and Phil Blunsom. 2015. Teaching machines to read and comprehend. In Advances in Neural Information Processing Systems, pages 1693– 1701.
  6. Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long short-term memory. Neural computation, 9(8):1735–1780. find
  7. Diederik P Kingma and Jimmy Ba. 2015. Adam: A method for stochastic optimization. In Proceedings of the International Conference on Learning Representations. find
  8. Thomas N Kipf and Max Welling. 2017. Semi-supervised classification with graph convolutional networks. In Proceedings of the International Conference on Learning Representations.
  9. Xiaodong Liu, Yelong Shen, Kevin Duh, and Jianfeng Gao. 2017. Stochastic answer networks for machine reading comprehension. arXiv preprint arXiv:1712.03556. find
  10. Christopher Manning, Mihai Surdeanu, John Bauer, Jenny Finkel, Steven Bethard, and David McClosky. 2014. The stanford corenlp natural language processing toolkit. In Proceedings of 52nd annual meeting of the association for computational linguistics: system demonstrations, pages 55–60.
  11. Sewon Min, Victor Zhong, Richard Socher, and Caiming Xiong. 2018. Efficient and robust question answering from minimal context over documents. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), volume 1, pages 1725–1735.
  12. Tsendsuren Munkhdalai and Hong Yu. 2017. Reasoning with memory augmented neural networks for language comprehension. In Proceedings of the International Conference on Learning Representations. find
  13. Takeshi Onishi, Hai Wang, Mohit Bansal, Kevin Gimpel, and David McAllester. 2016. Who did what: A large-scale person-centered cloze dataset. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 2230– 2235.
  14. Pranav Rajpurkar, Robin Jia, and Percy Liang. 2018. Know what you dont know: Unanswerable questions for squad. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), volume 2, pages 784–789.
  15. Pranav Rajpurkar, Jian Zhang, Konstantin Lopyrev, and Percy Liang. 2016. Squad: 100,000+ questions for machine comprehension of text. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 2383–2392.
  16. Erik F Tjong Kim Sang and Fien De Meulder. 2003. Introduction to the conll-2003 shared task: Language-independent named entity recognition. In Proceedings of the Seventh Conference on Natural Language Learning at HLT-NAACL 2003.
  17. Minjoon Seo, Aniruddha Kembhavi, Ali Farhadi, and Hannaneh Hajishirzi. 2016. Bidirectional attention flow for machine comprehension. In Proceedings of the International Conference on Learning Representations.
  18. Linfeng Song, Zhiguo Wang, Mo Yu, Yue Zhang, Radu Florian, and Daniel Gildea. 2018a. Exploring graph-structured passage representation for multi-hop reading comprehension with graph neural networks. arXiv preprint arXiv:1809.02040. — [arXiv:1809.02040](https://arxiv.org/abs/1809.02040) find
  19. Linfeng Song, Yue Zhang, Zhiguo Wang, and Daniel Gildea. 2018b. A graph-to-sequence model for amr-to-text generation. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), volume 1, pages 1616–1626.
  20. Sainbayar Sukhbaatar, Jason Weston, Rob Fergus, et al. 2015. End-to-end memory networks. In Advances in neural information processing systems, pages 2440–2448. find
  21. Alon Talmor and Jonathan Berant. 2018. The web as a knowledge-base for answering complex questions. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers), volume 1, pages 641–651.
  22. Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio. 2018. Graph attention networks. In Proceedings of the International Conference on Learning Representations. find
  23. Wenhui Wang, Nan Yang, Furu Wei, Baobao Chang, and Ming Zhou. 2017. Gated self-matching networks for reading comprehension and question answering. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), volume 1, pages 189–198.
  24. Johannes Welbl, Pontus Stenetorp, and Sebastian Riedel. 2018. Constructing datasets for multi-hop reading comprehension across documents. Transactions of the Association of Computational Linguistics, 6:287–302.
  25. Zhilin Yang, Peng Qi, Saizheng Zhang, Yoshua Bengio, William Cohen, Ruslan Salakhutdinov, and Christopher D Manning. 2018. Hotpotqa: A dataset for diverse, explainable multi-hop question answering. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 2369–…
  26. Mantong Zhou, Minlie Huang, and Xiaoyan Zhu. 2018. An interpretable reasoning network for multi-relation question answering. In Proceedings of the 27th International Conference on Computational Linguistics, pages 2010–2022. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/qiu-2019-dynamically-fused-graph-network-multi/