Detecting Compositionality in Multi-Word Expressions

What it is about

The paper decides whether a multi-word expression (MWE) is compositional, that is, whether its meaning follows from its parts. The focus is compound nominals, proper names and adjective-noun constructions. For each MWE the method collects the context words of the expression and of its parts from a corpus, builds a graph of their co-occurrence, and clusters it to induce senses. It keeps only the major senses. The only language-dependent parts are a part-of-speech tagger and a parser. Because no compositionality test set existed, the authors built one from the 52,217 MWEs of WordNet 3.0, with a semi-automatic step to aid annotation. The best setting reaches 68.42% accuracy for detecting non-compositional MWEs, above a one-cluster baseline. The similarity measure Jsn beats Jc in every setting.

Ideas

• I1. A cheap check in onto/src/wsd.rs before any sense induction: for each EWT compound, measure the overlap between the context words of the whole and those of its head. Label it compositional when the overlap is high. Measure: accuracy on a hand-checked sample of 200 WordNet MWEs, two identical runs.
• I2. Put compositionality into en/src/lexgen.rs as a flag on each compound row, filled only from the one-cluster baseline first. Measure: the number of compound readings the noun-noun rule adds on EWT, with and without the flag.
• I3. Build a small compositionality set from WordNet multi-word lexemes the way the paper does, as a test set only. The WordNet data is open, but the judgments are the authors' and must be checked by hand. Measure: agreement with a second annotator on 50 items.

For Mova

The chapter is 19. The lexicon: vocabulary, multiword expressions and register. The code is onto/src/wsd.rs (the context step) and en/src/lexgen.rs (the flag). Sense induction is not CPU-cheap in its full form, so I1 is the first step to measure. The paper's numbers are for WordNet MWEs and not for EWT.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.3115/1667583.1667605

@inproceedings{korkontzelos2009detecting,
  title = {{Detecting Compositionality in Multi-Word Expressions}},
  author = {Korkontzelos, Ioannis and Manandhar, Suresh},
  booktitle = {Proceedings of the ACL-IJCNLP 2009 Conference Short Papers},
  year = {2009},
  doi = {10.3115/1667583.1667605},
  url = {https://doi.org/10.3115/1667583.1667605}
}

References

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

1. E. Agirre and A. Soroa. 2007. Semeval-2007, task 02: Evaluating WSI and discrimination systems. In proceedings of SemEval-2007. ACL. find
2. E. Agirre, D. Martı́nez, O. de Lacalle, and A. Soroa. 2006. Two graph-based algorithms for state-of-the-art WSD. In proceedings of EMNLP-2006. ACL. find
3. T. Baldwin, C. Bannard, T. Tanaka, and D. Widdows. 2003. An empirical model of MWE decomposability. In proceedings of the MWE workshop. ACL. find
4. T. Baldwin. 2006. Compositionality and MWEs: Six of one, half a dozen of the other? In proceedings of the MWE workshop. ACL. find
5. C. Bannard, T. Baldwin, and A. Lascarides. 2003. A statistical approach to the semantics of verb-particles. In proceedings of the MWE workshop. ACL. find
6. C. Biemann. 2006. Chinese whispers - an efficient graph clustering algorithm and its application to NLP problems. In proceedings of TextGraphs. ACL. find
7. T. Brants and A. Franz. 2006. Web 1t 5-gram corpus, version 1. Technical report, Google Research. find
8. P. Cook, A. Fazly, and S. Stevenson. 2008. The VNC-Tokens Dataset. In proceedings of the MWE workshop. ACL. find
9. T. Dunning. 1993. Accurate methods for the statistics of surprise and coincidence. Computational Linguistics, 19(1):61–74. find
10. G. Katz and E. Giesbrecht. 2006. Automatic identification of non-compositional MWEs using latent semantic analysis. In proceedings of the MWE workshop. ACL. find
11. I. P. Klapaftis and S. Manandhar. 2008. WSI using graphs of collocations. In proceedings of ECAI-2008. find
12. D. Klein and C. Manning. 2003. Fast exact inference with a factored model for natural language parsing. In proceedings of NIPS 15. MIT Press. find
13. I. Korkontzelos, I. Klapaftis, and S. Manandhar. 2009. Graph connectivity measures for unsupervised parameter tuning of graph-based sense induction systems. In proceedings of the UMSLLS Workshop, NAACL HLT 2009. find
14. L. Lee. 1999. Measures of distributional similarity. In proceedings of ACL. find
15. D. McCarthy, B. Keller, and J. Carroll. 2003. Detecting a continuum of compositionality in phrasal verbs. In proceedings of the MWE workshop. ACL. find
16. G. A. Miller. 1995. WordNet: a lexical database for English. ACM, 38(11):39–41. find our card
17. D. Yarowsky. 1995. Unsupervised WSD rivaling supervised methods. In proceedings of ACL. find

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/aclbulk-2009-korkontzelos-detecting-compositionality-multi-word-expressions/