SEM-I Rational MT: Enriching Deep Grammars with a Semantic Interface for Scalable Machine Translation

What it is about

The paper describes the LOGON system, which translates Norwegian to English using minimal recursion semantics in a transfer module, with two existing broad-coverage deep grammars. Each grammar publishes a semantic interface, called SEM-I, which lists its elementary predications and captures classes of lexical regularity. The interface lets the transfer work without knowing grammar internals, keeps ambiguity underspecified rather than expanding every reading, and uses type hierarchies so that rules can be general. The authors argue that such an interface is what makes a deep-grammar translation engine scale, because it gives a complete and reliable list of what each grammar can produce. The paper is a system description and gives no translation score in this text.

Ideas

• I1. Publish an interface table for each reader: the list of meta concepts it can produce, with their tags and relations. The encoder can then check coverage before it runs, and the gaps for each reader (for example the Hebrew words with no concept) can be counted per verse.
• I2. A type hierarchy for concepts: when a child concept has no chunk in the encoder, fall back to its parent (for example a second-person pronoun form to the general pronoun). This generalises the back-off chain in profile.rs from contexts to concepts, and the fallback counts can be logged.
• I3. Coverage as a deterministic gate with a negative control: remove one concept from the interface, run the gate, and check that it goes red. Run it twice to show the report is identical.

For Mova

The code is tools/translate (meta.rs and hebrew.rs for the concept sets, profile.rs for the chunks). The interface idea answers open problem 7 on the Hebrew reader's coverage, and the hierarchy idea answers the encoder's unseen-concept case, where Mova now falls back to a bare form from morph.

Topics: Machine translation

Cite

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

@inproceedings{flickinger2005sem,
  title = {{SEM-I Rational MT: Enriching Deep Grammars with a Semantic Interface for Scalable Machine Translation}},
  author = {Flickinger, Dan and Lønning, Jan Tore and Dyvik, Helge and Oepen, Stephan and Bond, Francis},
  booktitle = {Proceedings of Machine Translation Summit X: Papers},
  year = {2005},
  url = {https://aclanthology.org/2005.mtsummit-papers.22/}
}

References

14 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

1. Bos, J. (1995). Predicate logic unplugged. In Proceedings of the Tenth Amsterdam Colloquium (pp. 133 – 142). Amsterdam, The Netherlands. find
2. Bunt, H. (2003). Underspecification in semantic representations: Which technique for what purpose? In H. Bunt, I. van der Sluis, & R. Morante (Eds.), Proceedings of the Fifth International Workshop on Computational Semantics IWCS-5 (pp. 37 – 54). Tilburg, The Netherlands. find
3. Butt, M., Dyvik, H., King, T. H., Masuichi, H., & Rohrer, C. (2002). The parallel grammar project. In Proc. of 19th Int. Conference on Computational Linguistics, Coling 2002. Taipei, Taiwan. find
4. Copestake, A., & Flickinger, D. (2003). The semantic interface to the ERG and other LinGO grammars. DeepThought draft tech report, University of Cambridge, UK. find
5. Copestake, A., Flickinger, D., Malouf, R., Riehemann, S., & Sag, I. (1995). Translation using Minimal Recursion Semantics. In Proceedings of the Sixth International Conference on Theoretical and Methodological Issues in Machine Translation (pp. 15–32). Leuven, Belgium. find our card
7. Dyvik, H. (2003). ParGram. Developing parallel grammars. ELSNews, 12 (2), 12 – 14. find
8. Genabith, J. van, & Crouch, R. (1997). Interpreting f-structures as UDRSs. In Proc. of the 35th Assoc. for Computational Linguistics and the 7th European ACL (pp. 402 – 409). Madrid, Spain. find
9. Ikehara, S., Shirai, S., & Bond, F. (1996). Approaches to disambiguation in ALT-J/E. In Int. seminar on multimodal interactive disambiguation: Middim-96 (pp. 107–117). Grenoble. find
10. Mahesh, K., Nirenburg, S., Beale, S., Viegas, E., Raskin, V., & Onyshkevych, B. (1997). Word sense disambiguation: Why statistics when you have these numbers? In Proc. of 7th Int. Conf. on Theoretical and Methodological Issues in Machine Translation: TMI-97 (pp. 151–159). Santa Fe. find
11. Oepen, S., Dyvik, H., Lønning, J. T., Velldal, E., Beermann, D., Carroll, J., Flickinger, D., Hellan, L., Johannessen, J. B., Meurer, P., Nordgård, T., & Rosén, V. (2004). Som å kapp-ete med trollet? Towards MRS-based Norwegian – English Machine Translation. In Proc. of 10th Int. Conf. on Theoretical and Methodological Issues in Machine Translation (pp. 11 – 20). Baltimore, MD. find
12. Pelletier, F. J. (Ed.). (1979). Mass terms: Some philosophical problems. Dordrecht: Reidel. find
13. Reyle, U. (1993). Dealing with ambiguity by underspecification. Construction, representation, and deduction. Journal of Semantics, 10, 123 – 179. find
14. Velldal, E., Oepen, S., & Flickinger, D. (2004). Paraphrasing treebanks for stochastic realization ranking. In Proc. of 3rd Workshop on Treebanks and Linguistic Theories. Tübingen, Germany. find
15. Wahlster, W. (Ed.). (2000). Verbmobil. Foundations of speech-to-speech translation. Berlin: Springer. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/flickinger-2005-sem-i-rational-mt-enriching/