Elamur Research⌕
☰
home / papers / 1988 / Grammatical Category Disambiguation by Statistical Optimization

Grammatical Category Disambiguation by Statistical Optimization

Paper · 1988 · Steven J. DeRose

Summary

The paper addresses part-of-speech disambiguation for English. Many word forms can belong to several grammatical categories, and a tagger has to choose one without relying on syntactic or semantic analysis. DeRose reviews earlier systems: the rule-based approach of Klein and Simmons, TAGGIT, and CLAWS, the statistical tagger built for the LOB Corpus. CLAWS scores tag sequences using probabilities of adjacent tag pairs and is accurate. However, it needs non-polynomial time and space and depends on hand-added extras whose individual effects are unknown. The author's algorithm, VOLSUNGA, keeps the tag-transition statistics, removes most of those extras and redesigns the search for the best tag path so that it runs in linear time. Its dictionary and probability tables come from the entire Brown Corpus, and it was tested on that same corpus of about one million words. It reached roughly 96% accuracy, close to CLAWS. Because the training and test data are the same, this figure is somewhat optimistic. The result shows that a simple, fast statistical tagger can serve as a practical front end for parsers and other NLP systems.

Key points

  • About 11.5% of vocabulary types and over 40% of running tokens in the Brown Corpus are categorially ambiguous. "still" is the extreme case, with 7 possible tags.
  • Earlier systems:
    • TAGGIT used about 3,300 context-frame rules and an 86-tag set, and tagged about 77% of the Brown Corpus correctly.
    • CLAWS used about 130 tags and was calibrated on about 200,000 words. It reported 96–97% accuracy, but idiom pre-tagging alone contributed about 3% of that.
  • VOLSUNGA runs in linear time, which the author argues is optimal, while CLAWS has exponential cost. VOLSUNGA also tolerates ungrammatical input.
  • Overall accuracy was 96.04% on about 1.14M tokens, including punctuation. Informative prose scored 96.20% and imaginative prose 95.57%.
  • By genre, accuracy ranged from 95.34% (E: Skills/Hobbies) to 96.66% (H: Miscellaneous).

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Brown UniversityProvidence · United States58 works · 25,310 citations

📚 Shelves 4 Computer Science · 9,086 works Artificial Intelligence · 8,889 works Arts and Humanities · 9,296 works Language and Linguistics · 9,206 works

Cite

Cite as Steven J. DeRose. 1988. Grammatical Category Disambiguation by Statistical Optimization. Computational Linguistics.

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

@article{derose1988grammatical,
  title = {{Grammatical Category Disambiguation by Statistical Optimization}},
  author = {Steven J. DeRose},
  journal = {Computational Linguistics},
  year = {1988},
  doi = {10.5555/49084.49087},
  url = {https://doi.org/10.5555/49084.49087}
}

References

15 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Beale, Andrew David. 1985 Grammatical Analysis by Computer of the Lancaster-Oslo/Bergen (LOB) Corpus of British English Texts. Proceedings of the 23rd Annual Meeting of the Association for Computational Linguistics. University of Chicago Press, Chicago, Illinois: 293-298. find
  2. Blackwell, Sue A. 1985 A Survey of Computer-Based English Language Research. ICAME News 9: 3-28. (Available from the Norwegian Computing Centre for the Humanities, Harald Harfa-gres gate 31, P.O. Box 53, N-5014 Bergen University, Norway.) find
  3. Booth, B. M. 1985 Revising CLAWS. ICAME News 9:29-35. find
  4. Dano, Sven. 1975 Nonlinear and Dynamic Programming. Springer-Verlag, New York. find
  5. Dreyfus, Stuart E. and Law, Averill, M. 1977 The Art and Theory of Dynamic Programming. Academic Press, New York. find
  6. Francis, W. Nelson and Kucera, Henry. 1979 Manual of Information to Accompany A Standard Corpus of Present-Day Edited American English, for Use with Digital Computers ("Revised and Amplified" edition). Department of Linguistics, Brown University, Providence, Rhode Island. find
  7. Francis, W. Nelson and Kucera, Henry. 1982 Frequency Analysis of English Usage: Lexicon and Grammar. Houghton-Mifflin Company, Boston, Massachusetts. find
  8. Greene, Barbara B. and Rubin, Gerald M. 1971 Automated Grammatical Tagging of English. Department of Linguistics, Brown University, Providence, Rhode Island. find
  9. Hirst, Graeme. 1983 Semantic Interpretation Against Ambiguity. Ph.D. dissertation., Department of Computer Science, Brown University, Providence, Rhode Island. find
  10. Klein, S. and Simmons, R. F. 1963 A Computational Approach to Grammatical Coding of English Words. JACM 10: 334-47. find
  11. Kucera, Henry and Francis, W. Nelson. 1967 Computational Analysis of Present-Day American English. Brown University Press, Providence, Rhode Island. find
  12. Leech, Geoffrey; Garside, Roger; and Atwell, Erik. 1983 The Automatic Grammatical Tagging of the LOB Corpus. ICAME News 7: 13-33. find
  13. Marshall, Ian. 1983 Choice of Grammatical Word-Class Without Global Syntactic Analysis: Tagging Words in the LOB Corpus. Computers in the Humanities 17: 139-150. find
  14. Ryder, Joan and Walker, Edward C. T. 1982 Two Mechanisms of Lexical Ambiguity. In Mehler, Jacques; Walker, Edward C.T.; and Garret, Merrill, Eds., Perspectives on Mental Representation. Lawrence Erlbaum Associates, Hillsdale, New Jersey: 134-149. find
  15. Shannon, Claude E. 1951 Prediction and Entropy of Printed English. Bell System Technical Journal 30: 50–64. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/derose-1988-grammatical-category-disambiguation-statistical-optimization/