Elamur Research
☰
home / papers / 2017 / Bilateral Multi-Perspective Matching for Natural Language Sentences

Bilateral Multi-Perspective Matching for Natural Language Sentences

Written for the Research library (claude-opus-5-5) from the paper text, in our own words: the license "arXiv — non-exclusive license to distribute" does not allow redistributing the text itself. Original — arxiv.org/pdf/1702.03814

Summary

The paper addresses natural language sentence matching (NLSM): deciding how two sentences relate, a problem underlying tasks such as recognizing paraphrases, judging the inferential relationship between a premise and a hypothesis, and choosing the sentence that answers a question. The authors argue that earlier "matching-aggregation" systems had two weaknesses. They compared sentences in only one direction, or they used only one level of granularity, such as word-to-word. Their BiMPM model addresses both. Each word is represented by combining a pretrained vector (GloVe or word2vec) with a character-level LSTM embedding. A shared BiLSTM then adds context. The core layer compares every time-step of one sentence with the entire other sentence, in both directions. It does this through a multi-perspective cosine function, where each perspective is a learned weighting of the dimensions, and four matching strategies. A second BiLSTM condenses the results into a fixed-length vector, which a two-layer feed-forward network with softmax classifies. The authors report state-of-the-art or competitive results on all three tasks, which shows the architecture works as a general tool for sentence-pair problems.

Key points

  • The matching operation uses l learned perspectives, each a row of a weight matrix W ∈ ℝ^{l×d}. Its four strategies are Full-Matching, Maxpooling-Matching, Attentive-Matching and Max-Attentive-Matching.
  • On SNLI, matching the hypothesis against the premise (P ← Q) clearly beat the opposite direction. The full bilateral model improved further. An ensemble of four BiMPM models with different random seeds outperformed the ensemble of Chen et al. (2016).
  • For paraphrase identification, BiMPM beat a re-implemented L.D.C. model by more than 2 points. Multi-Perspective-CNN/LSTM baselines also clearly beat their Siamese-CNN/LSTM counterparts.
  • For answer selection on TREC-QA (clean version), BiMPM reached 0.802 MAP and 0.875 MRR, roughly tied with the best prior system (0.801 / 0.877).
  • On WikiQA, BiMPM reached 0.718 MAP and 0.731 MRR, below the best listed baseline (Wang and Jiang, 2016: 0.743 / 0.755). The authors nonetheless describe this as on par.

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

Fields: Other

Cited by

184 papers of our library cite it (of the 140673 whose bibliographies we read); 3 of them have a page here, 181 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.1702.03814 · arXiv 1702.03814

@misc{wang2017bilateral,
  title = {{Bilateral Multi-Perspective Matching for Natural Language Sentences}},
  author = {Wang, Zhiguo and Hamza, Wael and Florian, Radu},
  year = {2017},
  doi = {10.48550/arxiv.1702.03814},
  eprint = {1702.03814},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.1702.03814}
}

References

37 references, 7 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Samuel R Bowman, Gabor Angeli, Christopher Potts, and Christopher D Manning. A large annotated corpus for learning natural language inference. arXiv preprint arXiv:1508.05326, 2015. find
  2. Jane Bromley, James W. Bentz, Léon Bottou, Isabelle Guyon, Yann LeCun, Cliff Moore, Eduard Säckinger, and Roopak Shah. Signature verification using a ”siamese” time delay neural network. IJPRAI, 7(4):669– 688, 1993. find
  3. Qian Chen, Xiaodan Zhu, Zhenhua Ling, Si Wei, and Hui Jiang. Enhancing and combining sequential and tree lstm for natural language inference. arXiv preprint arXiv:1609.06038, 2016. find
  4. Jianpeng Cheng, Li Dong, and Mirella Lapata. Long short-term memory-networks for machine reading. arXiv preprint arXiv:1601.06733, 2016. find our card
  5. Hua He and Jimmy Lin. Pairwise word interaction modeling with deep neural networks for semantic similarity measurement. In NAACL, 2016. find
  6. Michael Heilman and Noah A Smith. Tree edit models for recognizing textual entailments, paraphrases, and answers to questions. In NAACL, 2010. find our card
  7. Sepp Hochreiter and Jürgen Schmidhuber. Long short-term memory. Neural computation, 9(8):1735–1780, 1997. find
  8. Diederik Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014. find
  9. Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. Deep learning. Nature, 521(7553):436– 444, 2015. find
  10. Pengfei Liu, Xipeng Qiu, and Xuanjing Huang. Modelling interaction of sentence pair with coupled-lstms. arXiv preprint arXiv:1605.05573, 2016. find
  11. Yang Liu, Chengjie Sun, Lei Lin, and Xiaolong Wang. Learning natural language inference using bidirectional lstm model and inner-attention. arXiv preprint arXiv:1605.09090, 2016. — [arXiv:1605.09090](https://arxiv.org/abs/1605.09090) find
  12. Yishu Miao, Lei Yu, and Phil Blunsom. Neural variational inference for text processing. In ICML, 2016. find
  13. Tomas Mikolov, Ilya Sutskever, Kai Chen, Greg S Corrado, and Jeff Dean. Distributed representations of words and phrases and their compositionality. In Advances in neural information processing systems, pages 3111–3119, 2013. find
  14. Lili Mou, Rui Men, Ge Li, Yan Xu, Lu Zhang, Rui Yan, and Zhi Jin. Natural language inference by tree-based convolution and heuristic matching. arXiv preprint arXiv:1512.08422, 2015. find our card
  15. Tsendsuren Munkhdalai and Hong Yu. Neural tree indexers for text understanding. arXiv preprint arXiv:1607.04492, 2016. find
  16. Ankur P Parikh, Oscar Täckström, Dipanjan Das, and Jakob Uszkoreit. A decomposable attention model for natural language inference. arXiv preprint arXiv:1606.01933, 2016. find our card
  17. Xi Peng, Junzhou Huang, Qiong Hu, Shaoting Zhang, Ahmed Elgammal, and Dimitris Metaxas. From circle to 3-sphere: Head pose estimation by instance parameterization. Computer Vision and Image Understanding, 136:92–102, 2015. find
  18. Xi Peng, Shaoting Zhang, Yu Yang, and Dimitris N Metaxas. Piefa: Personalized incremental and ensemble face alignment. In Proceedings of the IEEE International Conference on Computer Vision, pages 3880– 3888, 2015. find
  19. Xi Peng, Rogerio S Feris, Xiaoyu Wang, and Dimitris N Metaxas. A recurrent encoder-decoder network for sequential face alignment. In European Conference on Computer Vision, pages 38–56. Springer International Publishing, 2016. find
  20. Xi Peng, Shaoting Zhang, Yang Yu, and Dimitris N Metaxas. Toward personalized modeling: Incremental and ensemble alignment for sequential faces in the wild. International Journal of Computer Vision, pages 1–14, 2017. find
  21. Jeffrey Pennington, Richard Socher, and Christopher D Manning. Glove: Global vectors for word representation. In EMNLP, 2014. find
  22. Jinfeng Rao, Hua He, and Jimmy Lin. Noise-contrastive estimation for answer selection with deep neural networks. In CIKM, 2016. find
  23. Tim Rocktäschel, Edward Grefenstette, Karl Moritz Hermann, Tomáš Kočiskỳ, and Phil Blunsom. Reasoning about entailment with neural attention. arXiv preprint arXiv:1509.06664, 2015. find our card
  24. Cicero dos Santos, Ming Tan, Bing Xiang, and Bowen Zhou. Attentive pooling networks. arXiv preprint arXiv:1602.03609, 2016. — [arXiv:1602.03609](https://arxiv.org/abs/1602.03609) find
  25. Lei Sha, Baobao Chang, Zhifang Sui, and Sujian Li. Reading and thinking: Re-read lstm unit for textual entailment recognition. In COLING, 2016. find
  26. Ming Tan, Cicero dos Santos, Bing Xiang, and Bowen Zhou. Lstm-based deep learning models for non-factoid answer selection. arXiv preprint arXiv:1511.04108, 2015. — [arXiv:1511.04108](https://arxiv.org/abs/1511.04108) find
  27. Ivan Vendrov, Ryan Kiros, Sanja Fidler, and Raquel Urtasun. Order-embeddings of images and language. arXiv preprint arXiv:1511.06361, 2015. find
  28. Zhiguo Wang and Abraham Ittycheriah. Faq-based question answering via word alignment. arXiv preprint arXiv:1507.02628, 2015. — [arXiv:1507.02628](https://arxiv.org/abs/1507.02628) find
  29. Shuohang Wang and Jing Jiang. Learning natural language inference with lstm. arXiv preprint arXiv:1512.08849, 2015. find
  30. Shuohang Wang and Jing Jiang. A compare-aggregate model for matching text sequences. arXiv preprint arXiv:1611.01747, 2016. — [arXiv:1611.01747](https://arxiv.org/abs/1611.01747) find
  31. Mengqiu Wang, Noah A Smith, and Teruko Mitamura. What is the jeopardy model? a quasi-synchronous grammar for qa. In EMNLP, 2007. find
  32. Bingning Wang, Kang Liu, and Jun Zhao. Inner attention based recurrent neural networks for answer selection. In ACL, 2016. find
  33. Zhiguo Wang, Haitao Mi, Wael Hamza, and Radu Florian. Multi-perspective context matching for machine comprehension. arXiv preprint arXiv:1612.04211, 2016. — [arXiv:1612.04211](https://arxiv.org/abs/1612.04211) find
  34. Zhiguo Wang, Haitao Mi, and Abraham Ittycheriah. Semi-supervised clustering for short text via deep representation learning. In CoNLL, 2016. find
  35. Zhiguo Wang, Haitao Mi, and Abraham Ittycheriah. Sentence similarity learning by lexical decomposition and composition. In COLING, 2016. find
  36. Yi Yang, Wen-tau Yih, and Christopher Meek. Wikiqa: A challenge dataset for open-domain question answering. In EMNLP, 2015. find our card
  37. Wenpeng Yin, Hinrich Schütze, Bing Xiang, and Bowen Zhou. Abcnn: Attention-based convolutional neural network for modeling sentence pairs. arXiv preprint arXiv:1512.05193, 2015. find our card