The MATE meta-scheme for coreference in dialogues in multiple languages
Summary
Researchers who annotate "coreference" in spoken dialogue have no agreed standard. Existing schemes such as MUCCS, DRAMA, the Lancaster scheme and the Bruneseaux–Romary scheme were each built for particular goals. Several of them target written English and ignore dialogue phenomena such as disfluencies, references to a shared visual scene and language-specific forms like clitics. The authors also stress that coreference and anaphora are distinct notions.
For the MATE project's annotation workbench, they argue that a single universal standard is unrealistic. Instead they propose a meta-scheme: a catalogue of annotation options from which application designers select what they need.
The meta-scheme has three parts:
- a core layer for identity relations between NP-introduced discourse entities, comparable to MUCCS;
- three optional extensions, covering references to visual situations, bridging relations, and a wider range of anaphors and antecedents (incorporated clitics, empty elements, discourse deixis).
Annotations are written in XML and stored separately from a chunk-based base file. Following DRT, coreference is expressed through the same relations used for anaphora. The contribution is a design proposal; how reliable the extensions are remains to be tested.
Key points
- Reported reliability evidence (Poesio and Vieira, 1998): whether an NP is anaphoric, K = .76; antecedents of agreed anaphoric definites, about 95% agreement; identifying bridging references, only K = .24. MUCCS is the only scheme whose reliability has been extensively tested.
- Markup elements:
<de>marks discourse entities;<link>with one or more<anchor>children encodes relations, where<link>points to the anaphor and<anchor>to the antecedent;- several anchors mark ambiguity, which is frequent in speech.
- Objects in the visual scene are listed as
<ue>items inside<universe>elements. For MapTask, where the two speakers have different maps, the authors suggest a shared "COMMON" universe plus one universe per participant. - Examples come from English (TRAINS-style dialogues, MapTask), French and Spanish. The Spanish examples illustrate clitics, which are also relevant to Italian.
- The documentation shows how to derive a specific scheme from the meta-scheme and how to encode the DRAMA scheme with its markup.
Cited by
- Can predicate-argument relationships be extracted from UD trees?
- Scoring Coreference Chains with Split-Antecedent Anaphors
- A General-Purpose, Off-the-shelf Anaphora Resolution Module: Implementation and Preliminary Evaluation
- The MATE/GNOME Proposals for Anaphoric Annotation, Revisited
- The Reliability of Anaphoric Annotation, Reconsidered: Taking Ambiguity into Account
- Challenges to Evaluating the Generalization of Coreference Resolution Models: A Measurement Modeling Perspective
- Sense Anaphoric Pronouns: Am I One?
Cite
@inproceedings{poesio1999mate,
title = {{The MATE meta-scheme for coreference in dialogues in multiple languages}},
author = {Poesio, M. and Bruneseaux, F. and Romary, L.},
booktitle = {Towards Standards and Tools for Discourse Tagging},
year = {1999},
url = {https://aclanthology.org/W99-0309/}
}References
- J. Alexandersson, B. Buschbeck-Wolf, T. Fujinami, E. Maier, N. Reithinger, B. Schmitz, and M. Siegel. 1997. Dialogue acts in \\;ERBMOBIL-2. Verbmobil Report 204, DFKI. find
- J. Allen and M. Core. 1997. DAMSL: Dialogue act markup in several layers. Draft contribution for the Discourse Resource Initiative, October. find
- F. Bruneseaux and L. Romary. 1998. Documents pr~paratoires pour le codage de dialogues multimodaux suivant les directives de la TEL Available at h t t p : / / w w w . l o r i a . f r / ' r o m a r y / Documents/index. html. find
- J. Carletta, A. Isard, S. Isard, J. Kowtko, G. Doherty-Sneddon, and A. H. Anderson. 1997. find
- The reliability of a dialogue structure coding scheme. Computational Linguistics, 23(1):13-32. find
- H. H. Clark. 1977. Bridging. In P. N. Johnson-Laird and P. C. Wason, editors, Thinking: Readings in Cognitive Science. Cambridge University Press. find
- Discourse Resource Initiative. 1997. Standards for dialogue coding in natural language processing. Report no. 167, Dagstuhl-Seminar. find
- S. Fligelstone. 1992. Developing a scheme for annotating text to show anaphoric relations. In G. Leitner, editor, New directions in English language corpora, pages 153-170. de Gruyter. find
- U. Held and A. Mengel. 1999. A query language for research in phonetics. In ICPhS, Sall Francisco. find
- I. Helm. 1982. The Semantics of Definite and Indefinite Noun Phrases. PhD thesis, University of Massachusetts at Amherst, 1982. find
- L. Hirschman. 1997. MUC-7 coreference task definition, version 3.0. Available at h t t p ://www.muc. s a i l . c o m / p r o c e e d i n g s / co\_task, html. find our card
- H. Kamp. 1981. A theory of truth and semantic representation. In J. Groenendijk, T. Janssen, and M. Stokhof, Formal Methods in the Study of Language, 277-322. Mathematisch Centrum. find
- R. Passonneau. 1997. Instructions for applying discourse reference annotation for multiple applications (DRAMA). Unpublished manuscript. find
- V. Pirrelli and C. Soria. 1999. Morpho-syntax annotation scheme. Available on the Web, February. find
- M. Poesio and R. Vieira. 1998. A corpus-based investigation of definite description use. Computational Linguistics, 24(2):183-216, June. find
- R. Quirk and S. Greenbaum. 1973. A Unive.rsity Grammar of English. Longman. find
- K. van Deemter and R. Kibble. 1999. What is coreference, and what should coreference annotation be? In Proc. of the ACL Workshop on Coreference, Maryland. 1998. Definite Description Resolution in Unrestricted Texts. Ph.D. thesis, University of Edinburgh, Centre for Cognitive Science. find our card
- B. Webber. 1991. Structure and ostension in the interpretation of discourse deixis. Language and Cognitive Processes, 6(2):107-135. find