A Linguistic Resource for Discovering Event Structures and Resolving Event Coreference

Summary

The paper addresses a gap in existing resources. FrameNet and WordNet describe event relations only at the conceptual level. PropBank and VerbNet cover only predicate-argument structure. TimeBank and ACE are limited to temporal links or a narrow set of event types. The authors build an annotated corpus of event structures, meaning groups of events that interact within one situation. Events are marked following TimeML. Links between them use six relations, each with a defined temporal reading, and coreference between events is annotated both within and across documents.

The data comes from clusters of news articles on shared themes, averaging about seven parallel articles per theme. Annotation was done with Callisto and modified versions of Tango. The paper also outlines automatic discovery of event structures. First, an adapted LDA model clusters events into latent structures, evaluated against FrameNet-derived scenarios. Second, supervised classifiers label relations within each cluster. A case study shows that FrameNet scenarios capture only a small share of the events in a real story. The resource supports causal, intentional and temporal inference, including deriving event chronologies.

Key points

Six event relations: subevent, reason, purpose, enablement, precedence and related. Each maps to a temporal constraint; for example, a subevent's interval lies inside the interval of its composite event.
Event clustering: LDA models documents as mixtures of event structures, using the lda-c tool with 50 event structures. It was tested on TimeBank events and evaluated by coverage of FrameNet frame-relation scenarios.
Annotation pipeline: themed news articles are cleaned and split into sentences. Events are then tagged per TimeML in Callisto, and coreference and relations are added with Tango. A modified Tango handles cross-document document pairs that share a theme.
Case study on the September 2007 arrest of Diego Montoya: the FrameNet crime scenario covers only about 20% of the annotated events. It also misses lexical units such as resistance and, in this context, reverses the temporal order of accused and captured.
Chronologies: the temporal interpretations of the relations yield a strict partial order (chronology) over the events of one structure.

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 · Semantic role labeling

Cited by

23 papers of our library cite it (of the 140673 whose bibliographies we read); 10 of them have a page here, 13 not yet.

Cite

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

@inproceedings{bejan2008linguistic,
  title = {{A Linguistic Resource for Discovering Event Structures and Resolving Event Coreference}},
  author = {Bejan, Cosmin and Harabagiu, Sanda},
  booktitle = {Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC'08)},
  year = {2008},
  doi = {10.63317/3o24r2j33x9v},
  url = {https://doi.org/10.63317/3o24r2j33x9v}
}

References

10 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

Collin F. Baker, Charles J. Fillmore, and John B. Lowe. 1998. The Berkeley FrameNet project. In Proceedings of the COLING-ACL. find our card
Cosmin Adrian Bejan and Chris Hathaway. 2007. UTD-SRL: A Pipeline Architecture for Extracting Frame Semantic Structures. In Proceedings of the Fourth International Workshop on Semantic Evaluations (SemEval-2007), pages 460–463. find our card
Cosmin Adrian Bejan. 2008. Unsupervised Discovery of Event Scenarios from Texts. In Proceedings of the 21st Florida Artificial Intelligence Research Society International Conference (FLAIRS), Applied Natural Language Processing track. find
David Blei, Andrew Ng, and Michael Jordan. 2003. Latent Dirichlet Allocation. Journal of Machine Learning Research, pages 993–1022. find our card
Michael Collins. 1997. Three generative, lexicalized models for statistical parsing. In Proceedings of the ACL– EACL, pages 16–23. find
Christiane Fellbaum. 1998. WordNet: An Electronic Lexical Database. MIT Press. find
Charles J. Fillmore. 1982. Frame Semantics. In Linguistics in the Morning Calm. find
Jerry R. Hobbs. 1985. On the coherence and Structure of Discourse. In CSLI Technical Report 85–37. find
Paul Kingsbury, Martha Palmer, and Mitch Marcus. 2002. Adding Semantic Annotation to the Penn TreeBank. In Proceedings of Human Language Technology (HLT-2002). find
Karin Kipper, Hoa Trang Dang, and Martha Palmer. 2000. Class-Based Construction of a Verb Lexicon. In Pro-escorted find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/bejan-2008-linguistic-resource-discovering-event-structures/

📄 A Linguistic Resource for Discovering Event Structures and Resolving Event Coreference