Elamur Research⌕
☰
home / papers / 2013 / Chinese Event Coreference Resolution: Understanding the State of the Art

Chinese Event Coreference Resolution: Understanding the State of the Art

Summary

The paper asks why event coreference, the task of linking mentions of the same real-world event, remains hard, using Chinese as a test case. The authors build an ACE 2005-style resolver that follows the mention-pair approach of Soon et al. (2001). It learns with SVM light, relies on 32 features adapted from an English system, and groups mentions through closest-first clustering. It depends on six upstream modules: event detection and subtyping, attribute prediction, argument and role labeling, entity coreference, entity mention detection, and entity typing. All modules begin as gold-standard oracles. Each is then swapped, one at a time, for its machine-learned version. This shows how much each source of pipeline noise lowers coreference quality. A follow-up error analysis looks at mistakes that come from the resolution algorithm itself. Many of these stem from judging whether arguments are compatible, which can require world knowledge. Because almost all earlier published results were for English, the reported scores act as a reference baseline for Chinese. The study also points to which components most deserve future effort.

Key points

  • Data and evaluation: 633 Chinese documents from the ACE 2005 training corpus, with five-fold cross-validation. Scoring uses MUC, B3 and CEAFe after singleton mentions are removed. Stanford's Chinese NLP tools handle segmentation and parsing.
  • Upper bound: with all upstream modules set to oracle, the resolver scores MUC F 74.8 (R 80.4, P 70.0) and B3 F 83.8.
  • Weak attribute predictors: the classifiers for Polarity, Modality, Genericity and Tense barely outperform a majority-class baseline. Tense is especially hard in Chinese because verbs do not inflect for tense.
  • Feature design: the 32 features fall into five groups. Four groups draw on the outputs of upstream components 1–4 (event subtypes, attributes, arguments and roles, entity coreference chains). The fifth covers lexical and distance cues, such as same trigger or sentence distance.
  • Error types and proposed fixes:
    • Missed links occur when compatible arguments look different on the surface, such as Bosnia and Herzegovina vs. Sarajevo. Fixing these needs world knowledge.
    • False links occur when a key role clashes, such as Destination Sweden vs. Denmark. The authors suggest marking some roles of each event subtype as important and blocking coreference when those roles conflict.

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

Topics: Coreference resolution

Cited by

6 papers of our library cite it (of the 140673 whose bibliographies we read); 1 of them have a page here, 5 not yet.

Cite

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

@inproceedings{chen2013chinese,
  title = {{Chinese Event Coreference Resolution: Understanding the State of the Art}},
  author = {Chen, Chen and Ng, Vincent},
  booktitle = {Proceedings of the Sixth International Joint Conference on Natural Language Processing},
  year = {2013},
  url = {https://aclanthology.org/I13-1100/}
}

References

18 references, 12 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Amit Bagga and Breck Baldwin. 1998. Algorithms for scoring coreference chains. In Proc. of the LREC Workshop on Linguistic Coreference, page 563--566. find our card
  2. Cosmin Bejan and Sanda Harabagiu. 2010. Unsupervised event coreference resolution with rich linguistic features. In Proc. of the ACL, pages 1412--1422. find our card
  3. Zheng Chen and Heng Ji. 2009. Graph-based event coreference resolution. In Proc. of TextGraphs-4, pages 54--57. find our card
  4. Chen Chen and Vincent Ng. 2012a. Chinese noun phrase coreference resolution: Insights into the state of the art. In Proc. of COLING 2012: Posters Volume, pages 185--194. find
  5. Chen Chen and Vincent Ng. 2012b. Combining the best of two worlds: A hybrid approach to multilingual coreference resolution. In Proc. of EMNLP-CoNLL: Shared Task, pages 56--63. find our card
  6. Chen Chen and Vincent Ng. 2012c. Joint modeling for Chinese event extraction with rich linguistic features. In Proc. of COLING, pages 529--544. find
  7. Zheng Chen, Heng Ji, and Robert Haralick. 2009. A pairwise event coreference model, feature impact and evaluation for event coreference resolution. In Proc. of the RANLP Workshop on Events in Emerging Text Types, pages 17--22. find our card
  8. Bin Chen, Jian Su, Sinno Jialin Pan, and Chew Lim Tan. 2011. A unified event coreference resolution by integrating multiple resolvers. In Proc. of IJC-NLP, pages 102--110. find our card
  9. Kevin Humphreys, Robert Gaizauskas, and Saliha Azzam. 1997. Event coreference for information extraction. In Proc. of the ACL Workshop on Operational Factors in Practical, Robust Anaphora Resolution for Unrestricted Texts, pages 75--81. find our card
  10. Thorsten Joachims. 1999. Making large-scale SVM learning practical. In Advances in Kernel Methods - Support Vector Learning, pages 44--56. MIT Press. find
  11. Heeyoung Lee, Marta Recasens, Angel Chang, Mihai Surdeanu, and Dan Jurafsky. 2012. Joint entity and event coreference resolution across documents. In Proc. of EMNLP-CoNLL, pages 489--500. find our card
  12. Xiaoqiang Luo. 2005. On coreference resolution performance metrics. In Proc. of HLT/EMNLP, pages 25--32. find our card
  13. Marta Recasens and Eduard Hovy. 2010. Coreference resolution across corpora: Languages, coding schemes, and preprocessing information. In Proc. of the ACL, pages 1423--1432. find our card
  14. Wee Meng Soon, Hwee Tou Ng, and Daniel Chung Yong Lim. 2001. A machine learning approach to coreference resolution of noun phrases. Computational Linguistics, 27(4):521--544. find our card
  15. Veselin Stoyanov, Nathan Gilbert, Claire Cardie, and Ellen Riloff. 2009. Conundrums in noun phrase coreference resolution: Making sense of the state-ofthe-art. In Proc. of ACL-IJCNLP, pages 656--664. find
  16. Ioannis Tsochantaridis, Thomas Hofmann, Thorsten Joachims, and Yasemin Altun. 2004. Support vector machine learning for interdependent and structured output spaces. In Proc. of ICML, pages 104--112. find
  17. Marc Vilain, John Burger, John Aberdeen, Dennis Connolly, and Lynette Hirschman. 1995. A model-theoretic coreference scoring scheme. In Proc. of MUC-6, pages 45--52. find our card
  18. Heike Zinmeister, Stefanie Dipper, and Melanie Seiss. 2012. Abstract pronominal anaphors and label nouns in German and English: Selected case studies and quantitative investigations. Translation: Corpora, Computation, Cognition, 2(1):47--80. find

↑ back to top

This page in Markdown →