Syntactic disambiguation using presupposition resolution
What it is about
Statistical parsers struggle with ambiguities where two readings are both common in training data, such as "The man saw the girl with the telescope". The authors propose choosing among the readings of an ambiguous sentence by how well each one fits the current discourse. Each reading becomes a DRS, and its presuppositions are resolved against the context. Two principles decide. The accommodation principle prefers the reading that needs the fewest added conditions. The presuppositional-weight principle, applied when two readings are both fully resolved, prefers the one with more presupposed conditions (for example, 4 conditions against 2 for the information-structure reading of "The cat chased the dog" in an answer to "What chased the dog?"). The method is used in the bilingual Te Kaitito dialogue system. The paper gives no accuracy numbers.
Ideas
- I1.
world/src/qud.rsandworld/src/qa.rs: when the parser returns several analyses, score each analysis by how many of its presupposed conditions the current world already satisfies, and keep the highest. Use integer counts and BTreeMap ordering so ties are deterministic. - I2. Add the accommodation count as a penalty: an analysis that needs a new entity in the world costs one unit per entity. Report both scores, so the choice is explained.
- I3. Test on a list of ambiguous school sentences with a short context each, comparing the chosen reading with the one a teacher would give; run twice.
For Mova
The chapter is 29. Ellipsis, word order and information structure (answers and focus) and 33. Implicature and presupposition. The code is world/src/qud.rs and world/src/qa.rs for the question context, and the parse candidates from en/src/parse.rs (or gparse). Measure: agreement with the intended reading on a written list of ambiguous sentences with contexts, both scores reported, run twice.
Cite
@inproceedings{knott2003syntactic,
title = {{Syntactic disambiguation using presupposition resolution}},
author = {Knott, Alistair and Vlugter, Peter},
booktitle = {Proceedings of the Australasian Language Technology Workshop 2003},
year = {2003},
url = {https://aclanthology.org/U03-1006/}
}References
- J Bos. 2001. DORIS 2001: Underspecification, resolution and inference for discourse representation structures. In Proceedings of ICoS-3: Inference in Computational Semantics. find
- M Collins. 1996. A new statistical parser based on bigram lexical dependencies. In Proceedings of the 34th Meeting of the ACL, Santa Cruz. find
- A Copestake, D Flickinger, I Sag, and C Pollard. 1999. Minimal Recursion Semantics: An introduction. Manuscript, CSLI, Stanford University. find
- A Copestake. 2000. The (new) LKB system. CSLI, Stanford University. find
- S de Jager, A Knott, and I Bayard. 2002. A DRT-based framework for presuppositions in dialogue management. In Proceedings of the 6th workshop on the semantics and pragmatics of dialogue (EDILOG 2002), Edinburgh. find
- F Ferreira and C Clifton. 1986. The independence of syntactic processing. Journal of Memory and Language, 25:348–368. ↗
- Joshua Goodman. 1998. Parsing Inside-Out. Ph.D. thesis, Harvard University. find
- J Hobbs, M Stickel, D Appelt, and P Martin. 1993. Interpretation as abduction. Artificial Intelligence, 63. find
- H Kamp and U Reyle. 1993. From discourse to logic. Kluwer Academic Publishers, Dordrecht. ↗
- A Knott, I Bayard, S de Jager, and N Wright. 2002. An architecture for bilingual and bidirectional nlp. In Proceedings of the 2nd Australasian Natural Language Processing Workshop (ANLP 2002). find
- David M. Magerman. 1995. Statistical decision-tree models for parsing. In Proc. of the 33 rd Annual Meeting of the Association for Computational Linguistics. Cambridge, MA, 26–30 Jun 1995. find
- M Steedman. 2000. The syntactic process. MIT Press/Bradford Books. ↗
- M Stone. 1998. Modality in dialogue: planning, pragmatics and computation. Ph.D. thesis, Institute for Research in Cognitive Science, University of Pennsylvania. find
- R Van der Sandt. 1992. Presupposition projection as anaphora resolution. Journal of Semantics, 9:333– 377. ↗
This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/knott-2003-syntactic-disambiguation-presupposition-resolution/