Anaphora Resolution in Machine Translation

Summary

The paper argues that pronoun resolution, long sidelined in MT because most systems work sentence by sentence, is often essential for correct output. Target languages may mark pronoun gender (German), choose different pronouns for human and non-human referents (Korean they), or prefer to drop the pronoun or replace it with a noun. In each case the system must know the antecedent. The authors also reject leaving anaphoric ambiguity unresolved. That approach gives no computational gain and becomes fragile as lexicons grow or new languages are added.

After reviewing earlier MT-oriented work, they describe an anaphora resolution model added to CAT2, a unification-based, Prolog-implemented transfer system. Anaphoric features of verb arguments are collected and passed up to the sentence node. This lets a pronoun in one sentence be matched against candidates in the previous sentence. Strict morphological and semantic agreement constraints are applied first, and preferential rules based on semantic roles then narrow any remaining ambiguity. The paper reports no quantitative evaluation. It shows only English–German examples that the model handles correctly. The main contribution is a concrete integration of discourse-level reasoning into a working MT formalism.

Key points

The motivating examples include the classic monkey/banana sentences, where German requires Affe, Banane, or es. In Korean, they must be resolved between programmers and programs to select a human or non-human form.
CAT2 was developed at IAI Saarbrücken as an offshoot of Eurotra and has been in development since 1987. It uses tree-to-tree transduction through an interface structure and keeps analysis and generation strictly monolingual.
The formalism has four rule types: b-rules and f-rules build trees and check their features, while t-rules and tf-rules handle transduction. It is implemented in SICStus Prolog.
Intersentential resolution works through a cand feature that percolates to the S node, with backtracking over candidate bindings. The demonstrated cases are council/commission/law sentences translated into German.
Planned extensions include a Bayesian center-tracking engine (Mitkov 94b), plus heuristic and domain modules and a referential expression filter.

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 · Machine translation

Topics: Coreference resolution · Machine translation

Cited by

13 papers of our library cite it (of the 140673 whose bibliographies we read); 4 of them have a page here, 9 not yet.

Cite

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

@inproceedings{mitkov1995anaphora,
  title = {{Anaphora Resolution in Machine Translation}},
  author = {Mitkov, Ruslan and Choi, Sung-Kwon and Sharp, Randall},
  booktitle = {Proceedings of the Sixth Conference on Theoretical and Methodological Issues in Machine Translation of Natural Languages},
  year = {1995},
  url = {https://aclanthology.org/1995.tmi-1.6/}
}

References

19 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

J. Carbonell, R. Brown. Anaphora resolution: a multi-strategy approach. Proceedings of COLING'88, Budapest, 1988 find our card
G. Hirst. Anaphora in natural language understanding. Berlin Springer Verlag, 1981 find
J. Hobbs. Resolving pronoun references. Lingua, Vol. 44, 1978 ↗ our card
J. Hutchins, H. Somers. An Introduction to Machine Translation, Academic Press, 1992 find
P. Isabelle, L. Bourbeau. TAUM-AVIATION: Its technical features and some experimental results, Computational Linguistics 11, 1985 find
R. Ingria, D. Stallard. A computational mechanism for pronominal reference. Proceedings of the 27th Annual Meeting of the ACL, Vancouver, Canada, 1989 find
R. Mitkov. An integrated model for anaphora resolution. Proceedings of the 15th International Conference on Computational Linguistics COLING'94, Kyoto, Japan, 1994 find our card
R. Mitkov. A new approach for tracking center. Proceedings of the International Conference “New Methods in Language Processing”, UMIST, Manchester, UK, 1994 find
R. Mitkov, H.K. Kim, H.K. Lee, K.S. Choi. The lexical transfer of pronominal anaphors in Machine Translation: the English-to-Korean case. Proceedings of the SEPLN'94 Conference, Cordoba, Spain, 1994 find
S. Nirenburg, J. Carbonell, M. Tomita, K. Goodman. Machine Translation: a knowledge-based approach, Morgan Kaufmann Publishers, 1992 find
S. Preuß, B. Schmitz, C. Hauenschild, C. Umbach. Anaphora Resolution in Machine Translation. In W. Ramm (ed): Studies in MT and NLP, Volume 6, EU Office, Luxembourg, 1994 find our card
E. Rich, S. LuperFoy. An architecture for anaphora resolution. Proceedings of the Second Conference on Applied Natural Language Processing, Austin, USA, 1988 find our card
H. Saggion, Ar. Carvalho. Anaphora resolution in a machine translation system. Proceedings of the International Conference “MT, 10 years on”, Cranfield, UK, 1994 find
R. Sharp. CAT2: Implementing a Formalism for Multi-Lingual MT. Proceedings of the TMI’88 Conference, Pittsburgh, USA, 1988. find
R. Sharp. CAT2 Reference Manual, Version 3.6. IAI, Saarbrücken, 1994 find
R. Sharp, O. Streiter. Applications in Multilingual Machine Translation. In: Proceedings of the Practical Applications of Prolog, Paris, 1995. find
C. Sidner. Focusing in the comprehension of definite anaphora. In B. Grosz, K. Jones, B. Webber (Eds): Readings in Natural Language Processing, Morgan Kaufmann Publishers, 1986 find
O. Streiter. Komplexe Disjunktion und erweiterter Kopf: Ein Kontrollmechanismus für die MÜ. Proceedings of Konvens’94, 1994, Vienna find
H. Wada. Discourse processing in MT: problems in pronominal translation. Proceedings of COLING’90, Helsinki, 1990 find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/papers/mitkov-1995-anaphora-resolution-machine-translation/

📄 Anaphora Resolution in Machine Translation