---
title: "Multi-Relational Script Learning for Discourse — Lee 2019"
description: "Most statistical script learning systems treat related events as vectors that should simply lie close together."
authors: ["I-Ta Lee", "Dan Goldwasser"]
year: 2019
venue: "Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics"
source: "https://aclanthology.org/P19-1413/"
doi: "10.18653/v1/p19-1413"
licence: "CC BY 4.0 (ACL Anthology)"
licence_url: "https://creativecommons.org/licenses/by/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/papers/lee-2019-multi-relational-script-learning-discourse/"
---
[Elamur Research](https://md.research.elamur.ai/) · [this page in HTML](https://research.elamur.ai/free/papers/lee-2019-multi-relational-script-learning-discourse/)

[home](https://md.research.elamur.ai/) / [papers](https://md.research.elamur.ai/free/papers/) / [2019](https://md.research.elamur.ai/free/papers/2019/page/3/) / Multi-Relational Script Learning for Discourse Relations

# Multi-Relational Script Learning for Discourse Relations

**Authors:** [I-Ta Lee](https://md.research.elamur.ai/free/authors/i-ta-lee/), [Dan Goldwasser](https://md.research.elamur.ai/free/authors/dan-goldwasser/) · **Year:** 2019 · **Venue:** Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics · **Licence:** [CC BY 4.0 (ACL Anthology)](https://creativecommons.org/licenses/by/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/P19-1413/) · [PDF](https://aclanthology.org/P19-1413.pdf) · [DOI 10.18653/v1/p19-1413](https://doi.org/10.18653/v1/p19-1413) · [OpenAlex](https://openalex.org/W2949875129)

**Citations:** 42 citations (openalex-snapshot · 2026-10-14) · [24](#cited-by) in our library, of the 140673 papers whose bibliographies we read

## Summary
Most statistical script learning systems treat related events as vectors that should simply lie close together. This captures co-occurrence but cannot tell whether one event causes, explains, contrasts with, or follows another. Lee and Goldwasser treat event embedding as a multi-relational problem instead.

Events are extracted with Stanford CoreNLP dependency parsing and coreference as predicate–subject–object tuples that keep full mention spans. Triplets linking these events come from three sources: shared coreferent entities, textual adjacency, and discourse markers that a rule-based annotator maps to PDTB senses. A GRU-based compositional encoder builds event vectors. These are trained with translation-based objectives adapted from knowledge-graph completion. EventTransE uses one shared space for events and relations, while EventTransR adds relation-specific projections.

The models outperform earlier work by a wide margin on MCNC and its sequential variants. They also beat an ELMo similarity baseline on relation-specific tests and improve an ELMo classifier for implicit discourse sense classification. The work shows that weak, marker-based supervision can yield embeddings sensitive to fine-grained commonsense relations.

## Key points
- The models cover 11 relation types: COREF NEXT, NEXT, and 9 PDTB-derived DISCOURSE NEXT senses, including Contrast, Reason, Result, Condition, Restatement, Conjunction, Instantiation, Synchrony and Asynchronous.
- There are two variants. EventTransE (TransE-style) has fewer parameters and converges faster, and it suits plain NEXT or COREF NEXT prediction. EventTransR (TransR-style) is stronger on relation-sensitive tasks.
- Intrinsic evaluations were:
  - choosing the next event under a given relation;
  - predicting which of the 9 discourse relations links two events;
  - triplet classification, using 500 positive and 500 negative samples per relation with relation-specific thresholds.
- In triplet classification, both models did especially well on Expansion relations (Instantiation, Restatement), and EventTransR scored highly on Asynchronous temporal relations.
- On the CoNLL 2016 implicit discourse sense task (15 senses), adding EventTransE to an attention-based ELMo baseline raised micro F1 on every split:

  | Split | ELMo | ELMo+EventTransE |
  |---|---|---|
  | Dev | 45.60 | 46.81 |
  | Test | 37.65 | 39.05 |
  | Blind | 36.72 | 38.35 |

  The combined model was the best system on the blind split. On the test split it remained below ecnucs (40.91).

*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: [Linguistics & languages](https://md.research.elamur.ai/free/shelves/linguistics/)

Topics: [Discourse analysis](https://md.research.elamur.ai/free/topics/discourse-analysis/)

## Cited by

Cited by

24 papers of our library cite it (of the 140673 whose bibliographies we read); 4 of them have a page here, 20 not yet.

- [Integrating Deep Event-Level and Script-Level Information for Script Event Prediction](https://md.research.elamur.ai/free/papers/bai-2021-integrating-deep-event-level-script/)
- [Discourse Analysis via Questions and Answers: Parsing Dependency Structures of Questions Under Discussion](https://md.research.elamur.ai/free/papers/ko-2023-discourse-analysis-questions-answers-parsing/)
- [Weakly-Supervised Modeling of Contextualized Event Embedding for Discourse Relations](https://md.research.elamur.ai/free/papers/lee-2020-weakly-supervised-modeling-contextualized-event/)
- [QADiscourse - Discourse Relations as QA Pairs: Representation, Crowdsourcing and Baselines](https://md.research.elamur.ai/free/papers/pyatkin-2020-qadiscourse-discourse-relations-as-qa/)

## Cite

The paper itself, not this page. BibTeX from the library's record of the work.

```bibtex
@inproceedings{lee2019multi,
  title = {{Multi-Relational Script Learning for Discourse Relations}},
  author = {Lee, I-Ta and Goldwasser, Dan},
  booktitle = {Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics},
  year = {2019},
  doi = {10.18653/v1/p19-1413},
  url = {https://doi.org/10.18653/v1/p19-1413}
}
```

- [← previous Multi-News: A Large-Scale Multi-Document Summarization Dataset and Abstractive Hierarchical Model](https://md.research.elamur.ai/free/papers/fabbri-2019-multi-news-large-scale-multi/)
- [next → Multi-Task Deep Neural Networks for Natural Language Understanding](https://md.research.elamur.ai/free/papers/liu-2019-multi-task-deep-neural-networks/)
- [↑ back to papers of 2019](https://md.research.elamur.ai/free/papers/2019/page/3/)

## References

30 references, 3 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, Alberto Garcia-Duran, Jason Weston, and Oksana Yakhnenko. 2013. Translating embeddings for modeling multi-relational data. In Advances in neural information processing systems, pages 2787–2795. [find](https://scholar.google.com/scholar?q=Antoine+Bordes%2C+Nicolas+Usunier%2C+Alberto+Garcia%2DDuran%2C+Jason+Weston%2C+and+Oksana+Yakhnenko%2E+2013%2E+Translating+embeddings+for+modeling+multi%2Drelational+data%2E+In+Advances+in+neural+information+processing)
2. Nathanael Chambers and Daniel Jurafsky. 2008. Unsupervised learning of narrative event chains. In ACL, volume 94305, pages 789–797. [find](https://scholar.google.com/scholar?q=Nathanael+Chambers+and+Daniel+Jurafsky%2E+2008%2E+Unsupervised+learning+of+narrative+event+chains%2E+In+ACL%2C+volume+94305%2C+pages+789%E2%80%93797%2E)
3. Junyoung Chung, Caglar Gulcehre, KyungHyun Cho, and Yoshua Bengio. 2014. Empirical evaluation of gated recurrent neural networks on sequence modeling. arXiv preprint arXiv:1412.3555. [find](https://scholar.google.com/scholar?q=Junyoung+Chung%2C+Caglar+Gulcehre%2C+KyungHyun+Cho%2C+and+Yoshua+Bengio%2E+2014%2E+Empirical+evaluation+of+gated+recurrent+neural+networks+on+sequence+modeling%2E+arXiv+preprint+arXiv%3A1412%2E3555%2E)
4. John Duchi, Elad Hazan, and Yoram Singer. 2011. Adaptive subgradient methods for online learning and stochastic optimization. JMLR, 12(Jul):2121– 2159. [find](https://scholar.google.com/scholar?q=John+Duchi%2C+Elad+Hazan%2C+and+Yoram+Singer%2E+2011%2E+Adaptive+subgradient+methods+for+online+learning+and+stochastic+optimization%2E+JMLR%2C+12%28Jul%29%3A2121%E2%80%93+2159%2E)
5. Mark Granroth-Wilding and Stephen Clark. 2016. What happens next? event prediction using a compositional neural network model. In AAAI, pages 2727–2733. [find](https://scholar.google.com/scholar?q=Mark+Granroth%2DWilding+and+Stephen+Clark%2E+2016%2E+What+happens+next%3F+event+prediction+using+a+compositional+neural+network+model%2E+In+AAAI%2C+pages+2727%E2%80%932733%2E)
6. Sepp Hochreiter and Jürgen Schmidhuber. 1997. Long short-term memory. Neural computation, 9(8):1735–1780. [find](https://scholar.google.com/scholar?q=Sepp+Hochreiter+and+Ju%CC%88rgen+Schmidhuber%2E+1997%2E+Long+short%2Dterm+memory%2E+Neural+computation%2C+9%288%29%3A1735%E2%80%931780%2E)
7. Bram Jans, Steven Bethard, Ivan Vulić, and Marie Francine Moens. 2012. Skip n-grams and ranking functions for predicting script events. In Proceedings of the 13th Conference of the European Chapter of the Association for Computational Linguistics, pages 336–344. Association for Computational Lingui… [find](https://scholar.google.com/scholar?q=Bram+Jans%2C+Steven+Bethard%2C+Ivan+Vulic%CC%81%2C+and+Marie+Francine+Moens%2E+2012%2E+Skip+n%2Dgrams+and+ranking+functions+for+predicting+script+events%2E+In+Proceedings+of+the+13th+Conference+of+the+European+Chapter+o)
8. Ryan Kiros, Yukun Zhu, Ruslan R Salakhutdinov, Richard Zemel, Raquel Urtasun, Antonio Torralba, and Sanja Fidler. 2015. Skip-thought vectors. In NIPS, pages 3294–3302. [find](https://scholar.google.com/scholar?q=Ryan+Kiros%2C+Yukun+Zhu%2C+Ruslan+R+Salakhutdinov%2C+Richard+Zemel%2C+Raquel+Urtasun%2C+Antonio+Torralba%2C+and+Sanja+Fidler%2E+2015%2E+Skip%2Dthought+vectors%2E+In+NIPS%2C+pages+3294%E2%80%933302%2E)
9. I-Ta Lee and Dan Goldwasser. 2018. Feel: Featured event embedding learning. AAAI, pages 4840–4847. [find](https://scholar.google.com/scholar?q=I%2DTa+Lee+and+Dan+Goldwasser%2E+2018%2E+Feel%3A+Featured+event+embedding+learning%2E+AAAI%2C+pages+4840%E2%80%934847%2E)
10. Zhongyang Li, Xiao Ding, and Ting Liu. 2018. Constructing narrative event evolutionary graph for script event prediction. arXiv preprint arXiv:1805.05081. [find](https://scholar.google.com/scholar?q=Zhongyang+Li%2C+Xiao+Ding%2C+and+Ting+Liu%2E+2018%2E+Constructing+narrative+event+evolutionary+graph+for+script+event+prediction%2E+arXiv+preprint+arXiv%3A1805%2E05081%2E)
11. Yankai Lin, Zhiyuan Liu, Maosong Sun, Yang Liu, and Xuan Zhu. 2015. Learning entity and relation embeddings for knowledge graph completion. In AAAI, volume 15, pages 2181–2187. [find](https://scholar.google.com/scholar?q=Yankai+Lin%2C+Zhiyuan+Liu%2C+Maosong+Sun%2C+Yang+Liu%2C+and+Xuan+Zhu%2E+2015%2E+Learning+entity+and+relation+embeddings+for+knowledge+graph+completion%2E+In+AAAI%2C+volume+15%2C+pages+2181%E2%80%932187%2E)
12. Nasrin Mostafazadeh, Michael Roth, Annie Louis, Nathanael Chambers, and James F Allen. 2017. [find](https://scholar.google.com/scholar?q=Nasrin+Mostafazadeh%2C+Michael+Roth%2C+Annie+Louis%2C+Nathanael+Chambers%2C+and+James+F+Allen%2E+2017%2E)
13. Ls-dsem 2017 shared task: The story cloze test. [find](https://scholar.google.com/scholar?q=Ls%2Ddsem+2017+shared+task%3A+The+story+cloze+test%2E)
14. Vinod Nair and Geoffrey E Hinton. 2010. Rectified linear units improve restricted boltzmann machines. In ICML, pages 807–814. [find](https://scholar.google.com/scholar?q=Vinod+Nair+and+Geoffrey+E+Hinton%2E+2010%2E+Rectified+linear+units+improve+restricted+boltzmann+machines%2E+In+ICML%2C+pages+807%E2%80%93814%2E)
15. Maximilian Nickel, Lorenzo Rosasco, Tomaso A Pog-gio, et al. 2016. Holographic embeddings of knowledge graphs. In AAAI, volume 2, pages 3–2. [find](https://scholar.google.com/scholar?q=Maximilian+Nickel%2C+Lorenzo+Rosasco%2C+Tomaso+A+Pog%2Dgio%2C+et+al%2E+2016%2E+Holographic+embeddings+of+knowledge+graphs%2E+In+AAAI%2C+volume+2%2C+pages+3%E2%80%932%2E)
16. Qiang Ning, Zhili Feng, Hao Wu, and Dan Roth. 2018. Joint reasoning for temporal and causal relations. In Proceedings of the 56th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), volume 1, pages 2278–2288. [find](https://scholar.google.com/scholar?q=Qiang+Ning%2C+Zhili+Feng%2C+Hao+Wu%2C+and+Dan+Roth%2E+2018%2E+Joint+reasoning+for+temporal+and+causal+relations%2E+In+Proceedings+of+the+56th+Annual+Meeting+of+the+Association+for+Computational+Linguistics+%28Volum)
17. Maria Leonor Pacheco, I-Ta Lee, Xiao Zhang, Abdullah Khan Zehady, Pranjal Daga, Di Jin, Ayush Paro-lia, and Dan Goldwasser. 2016. Adapting event embedding for implicit discourse relation recognition. Proceedings of the CoNLL-16 shared task, pages 136–142. [find](https://scholar.google.com/scholar?q=Maria+Leonor+Pacheco%2C+I%2DTa+Lee%2C+Xiao+Zhang%2C+Abdullah+Khan+Zehady%2C+Pranjal+Daga%2C+Di+Jin%2C+Ayush+Paro%2Dlia%2C+and+Dan+Goldwasser%2E+2016%2E+Adapting+event+embedding+for+implicit+discourse+relation+recognition%2E+) [our card](https://md.research.elamur.ai/free/papers/pacheco-2016-adapting-event-embedding-implicit-discourse/)
18. Robert Parker, David Graff, Junbo Kong, Ke Chen, and Kazuaki Maeda. 2011. English gigaword fifth edition ldc2011t07. dvd. Philadelphia: Linguistic Data Consortium. [find](https://scholar.google.com/scholar?q=Robert+Parker%2C+David+Graff%2C+Junbo+Kong%2C+Ke+Chen%2C+and+Kazuaki+Maeda%2E+2011%2E+English+gigaword+fifth+edition+ldc2011t07%2E+dvd%2E+Philadelphia%3A+Linguistic+Data+Consortium%2E)
19. Haoruo Peng and Dan Roth. 2016. Two discourse driven language models for semantics. arXiv preprint arXiv:1606.05679. [find](https://scholar.google.com/scholar?q=Haoruo+Peng+and+Dan+Roth%2E+2016%2E+Two+discourse+driven+language+models+for+semantics%2E+arXiv+preprint+arXiv%3A1606%2E05679%2E)
20. Matthew E Peters, Mark Neumann, Mohit Iyyer, Matt Gardner, Christopher Clark, Kenton Lee, and Luke Zettlemoyer. 2018. Deep contextualized word representations. arXiv preprint arXiv:1802.05365. [find](https://scholar.google.com/scholar?q=Matthew+E+Peters%2C+Mark+Neumann%2C+Mohit+Iyyer%2C+Matt+Gardner%2C+Christopher+Clark%2C+Kenton+Lee%2C+and+Luke+Zettlemoyer%2E+2018%2E+Deep+contextualized+word+representations%2E+arXiv+preprint+arXiv%3A1802%2E05365%2E)
21. Karl Pichotta and Raymond J Mooney. 2014. Statistical script learning with multi-argument events. In EACL, volume 14, pages 220–229. [find](https://scholar.google.com/scholar?q=Karl+Pichotta+and+Raymond+J+Mooney%2E+2014%2E+Statistical+script+learning+with+multi%2Dargument+events%2E+In+EACL%2C+volume+14%2C+pages+220%E2%80%93229%2E)
22. Karl Pichotta and Raymond J Mooney. 2016a. Learning statistical scripts with lstm recurrent neural networks. In AAAI. [find](https://scholar.google.com/scholar?q=Karl+Pichotta+and+Raymond+J+Mooney%2E+2016a%2E+Learning+statistical+scripts+with+lstm+recurrent+neural+networks%2E+In+AAAI%2E)
23. Karl Pichotta and Raymond J Mooney. 2016b. Using sentence-level lstm language models for script inference. arXiv preprint arXiv:1604.02993. [find](https://scholar.google.com/scholar?q=Karl+Pichotta+and+Raymond+J+Mooney%2E+2016b%2E+Using+sentence%2Dlevel+lstm+language+models+for+script+inference%2E+arXiv+preprint+arXiv%3A1604%2E02993%2E)
24. Rashmi Prasad, Eleni Miltsakaki, Nikhil Dinesh, Alan Lee, Aravind Joshi, Livio Robaldo, and Bonnie L Webber. 2007. The penn discourse treebank 2.0 annotation manual. [find](https://scholar.google.com/scholar?q=Rashmi+Prasad%2C+Eleni+Miltsakaki%2C+Nikhil+Dinesh%2C+Alan+Lee%2C+Aravind+Joshi%2C+Livio+Robaldo%2C+and+Bonnie+L+Webber%2E+2007%2E+The+penn+discourse+treebank+2%2E0+annotation+manual%2E)
25. Christopher D Manning, Mihai Surdeanu, John Bauer, Jenny Rose Finkel, Steven Bethard, and David Mc-Closky. 2014. The stanford corenlp natural language processing toolkit. In ACL (System Demonstrations), pages 55–60. [find](https://scholar.google.com/scholar?q=Christopher+D+Manning%2C+Mihai+Surdeanu%2C+John+Bauer%2C+Jenny+Rose+Finkel%2C+Steven+Bethard%2C+and+David+Mc%2DClosky%2E+2014%2E+The+stanford+corenlp+natural+language+processing+toolkit%2E+In+ACL+%28System+Demonstrations) [our card](https://md.research.elamur.ai/free/papers/manning-2014-stanford-corenlp-natural-language-processing/)
26. Hannah Rashkin, Maarten Sap, Emily Allaway, Noah A Smith, and Yejin Choi. 2018. Event2mind: Commonsense inference on events, intents, and reactions. arXiv preprint arXiv:1805.06939. [find](https://scholar.google.com/scholar?q=Hannah+Rashkin%2C+Maarten+Sap%2C+Emily+Allaway%2C+Noah+A+Smith%2C+and+Yejin+Choi%2E+2018%2E+Event2mind%3A+Commonsense+inference+on+events%2C+intents%2C+and+reactions%2E+arXiv+preprint+arXiv%3A1805%2E06939%2E)
27. Tomas Mikolov, Kai Chen, Greg Corrado, and Jeffrey Dean. 2013. Efficient estimation of word representations in vector space. arXiv preprint arXiv:1301.3781. [find](https://scholar.google.com/scholar?q=Tomas+Mikolov%2C+Kai+Chen%2C+Greg+Corrado%2C+and+Jeffrey+Dean%2E+2013%2E+Efficient+estimation+of+word+representations+in+vector+space%2E+arXiv+preprint+arXiv%3A1301%2E3781%2E)
28. Attapol Rutherford and Nianwen Xue. 2016. Robust non-explicit neural discourse parser in english and chinese. Proceedings of the CoNLL-16 shared task, pages 55–59. [find](https://scholar.google.com/scholar?q=Attapol+Rutherford+and+Nianwen+Xue%2E+2016%2E+Robust+non%2Dexplicit+neural+discourse+parser+in+english+and+chinese%2E+Proceedings+of+the+CoNLL%2D16+shared+task%2C+pages+55%E2%80%9359%2E) [our card](https://md.research.elamur.ai/free/papers/rutherford-2016-robust-non-explicit-neural-discourse/)
29. Maarten Sap, Ronan LeBras, Emily Allaway, Chandra Bhagavatula, Nicholas Lourie, Hannah Rashkin, Brendan Roof, Noah A Smith, and Yejin Choi. 2018. Atomic: An atlas of machine commonsense for if-then reasoning. arXiv preprint arXiv:1811.00146. [find](https://scholar.google.com/scholar?q=Maarten+Sap%2C+Ronan+LeBras%2C+Emily+Allaway%2C+Chandra+Bhagavatula%2C+Nicholas+Lourie%2C+Hannah+Rashkin%2C+Brendan+Roof%2C+Noah+A+Smith%2C+and+Yejin+Choi%2E+2018%2E+Atomic%3A+An+atlas+of+machine+commonsense+for+if%2Dthen+re)
30. Roger C Schank and Robert P Abelson. 1977. Scripts, plans, goals and understanding: An inquiry into human knowledge structures. [find](https://scholar.google.com/scholar?q=Roger+C+Schank+and+Robert+P+Abelson%2E+1977%2E+Scripts%2C+plans%2C+goals+and+understanding%3A+An+inquiry+into+human+knowledge+structures%2E)