Elamur Research⌕
☰
home / papers / 2009 / A Study on Linking Wikipedia Categories to Wordnet Synsets using Text Similarity

A Study on Linking Wikipedia Categories to Wordnet Synsets using Text Similarity

Paper · 2009 · Antonio Toral, Óscar Ferrández, Eneko Agirre et al.

Summary

The paper tackles a disambiguation step in building a Named Entity Repository from WordNet and Wikipedia. When polysemous WordNet nouns are matched to Wikipedia categories, it is unclear which sense a category refers to. An earlier approach used instances shared by both resources. It was perfectly precise but covered only 39% of cases. The authors therefore compare each WordNet gloss with the abstract of the matched category and pick the most similar sense.

They test several methods on a manually built, publicly released gold standard:

  • word overlap,
  • Semantic Vectors (random projection),
  • Personalised PageRank over the WordNet graph,
  • an in-house textual entailment (TE) system, applied in both directions and averaged.

They also try voting, unsupervised and supervised combinations of these methods.

The most-frequent-sense baseline is hard to beat at 64.7%. The best unsupervised result is 68%, from voting. A supervised TE model reaches 77.74%. The work shows that text-similarity techniques can resolve most ambiguous WordNet–Wikipedia links automatically, which matters for resource-linking projects such as YAGO.

Key points

  • Baselines: First Sense scores 64.7%. Word overlap scores 56.3%, rising to 62.7% once stop words are removed. Removing stop words also lifts Personalised PageRank from 61.8% to 64.3%.
  • The Semantic Vectors model was built from 117,598 WordNet 2.1 glosses and 2,179,275 Wikipedia abstracts (January 2008 dump). It scored lowest at 54.1%.
  • The TE system's accuracy depends heavily on training data:
    • trained on AVE 2007–08 and RTE-3 corpora: 52.8%;
    • without training: 64.7%;
    • supervised BayesNet with 10-fold cross-validation on the task data: 77.74%.
  • Combinations:
    • The oracle upper bound is 84.5%.
    • The best voting combination (PPR + TE + WO) reaches 68%.
    • No supervised combination beats TE alone; the best gets 77.24%. The authors attribute this to TE's internal features overlapping with the other methods.
  • Personalised PageRank used all WordNet 3.0 relations, including disambiguated glosses. It ran with damping 0.85 for 30 iterations, and these settings were not tuned for this task.

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

🎓University of the Basque CountryLeioa · Spain62 works · 16,683 citations🎓Dublin City UniversityDublin · Ireland73 works · 9,768 citations🎓University of AlicanteAlicante · Spain24 works · 1,251 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 19,503 worksArtificial Intelligence · 18,805 worksArts and Humanities · 17,959 worksLanguage and Linguistics · 17,789 works

Cite

Cite as Antonio Toral, Óscar Ferrández, Eneko Agirre and Rafael Muñoz. 2009. A Study on Linking Wikipedia Categories to Wordnet Synsets using Text Similarity. Proceedings of the International Conference RANLP-2009.

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

@inproceedings{toral2009study,
  title = {{A Study on Linking Wikipedia Categories to Wordnet Synsets using Text Similarity}},
  author = {Antonio Toral, Óscar Ferrández, Eneko Agirre and Rafael Muñoz},
  booktitle = {Proceedings of the International Conference RANLP-2009},
  year = {2009},
  url = {https://aclanthology.org/R09-1080/}
}

References

19 references, 4 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. E. Agirre, A. Soroa, E. Alfonseca, K. Hall, J. Kravalova, and M. Pasca. A study on similarity and relatedness using distributional and wordnet-based approaches. In Proceedings of NAACL, 2009.
  2. Ido Dagan and Oren Glickman. Probabilistic textual entailment: Generic appied modelling of language variability. In Proceedings of the PASCAL Workshop on Learning Methods for Text Understanding and Mining, Grenoble, France, 2004. find
  3. William B. Dolan and Chris Brockett. Automatically constructing a corpus of sentential paraphrases. In IWP2005, 2005.
  4. G. Erkan and Dragomir R. Radev. Lexrank: Graph-based centrality as salience in text summarization. Journal of Artificial Intelligence Research, 22:457–479, 2004. find
  5. Christian Fellbaum, editor. WordNet: An Electronic Lexical Database (ISBN: 0-262-06197-X). MIT Press, first edition, 1998. find
  6. Óscar Ferrández, Daniel Micol, Rafael Muñoz, and Manuel Palomar. A perspective-based approach for solving textual entailment recognition. In Proceedings of the ACL-PASCAL Workshop on Textual Entailment and Paraphrasing, pages 66– 71, Prague, June 2007. ACL. find
  7. Vasileios Hatzivassiloglou, Judith L. Klavans, Melissa L. Holcombe, Regina Barzilay, Min yen Kan, and Kathleen R. Mckeown. Simfinder: A flexible clustering tool for summarization. In NAACL Workshop on Automatic Summarization, pages 41–49, 2001. find
  8. T. H. Haveliwala. Topic-sensitive pagerank. In WWW ’02: Proceedings of the 11th international conference on World Wide Web, pages 517–526, New York, NY, USA, 2002. ACM.
  9. Aminul Islam and Diana Inkpen. Semantic text similarity using corpus-based word similarity and string similarity. ACM Trans. Knowl. Discov. Data, 2(2):1–25, July 2008. find
  10. Yuhua Li, David McLean, Zuhair A. Bandar, James D. O’Shea, and Keeley Crockett. Sentence similarity based on semantic nets and corpus statistics. IEEE Transactions on Knowledge and Data Engineering, 18(8):1138–1150, 2006. find
  11. Tao Liu and Jun Guo. Text similarity computing based on standard deviation. In De-Shuang Huang, Xiao-Ping Zhang, and Guang-Bin Huang, editors, ICIC (1), volume 3644 of Lecture Notes in Computer Science, pages 456–464. Springer, 2005. find
  12. Elena Lloret, Óscar Ferrández, Rafael Muñoz, and Manuel Palomar. A text summarization approach under the influence of textual entailment. In Natural Language Processing and Cognitive Science, Proceedings of the 5th International Workshop on Natural Language Processing and Cognitive Science (NLPCS… find
  13. Rada Mihalcea, Courtney Corley, and Carlo Strapparava. Corpus-based and knowledge-based measures of text semantic similarity. In In AAAI06, pages 775–780, 2006. find
  14. Óscar Ferrández, Rubén Izquierdo, Sergio Ferrández, and José Luis Vicedo. Addressing ontology-based question answering with collections of user queries. Information Processing and Management, In Press, Corrected Proof available online 31 October 2008, 2008. find
  15. Ted Pedersen. Computational approaches to measuring the similarity of short contexts : A review of applications and methods. CoRR, abs/0806.3787, 2008. — [arXiv:0806.3787](https://arxiv.org/abs/0806.3787) find
  16. Hinrich Schtitze. Automatic word sense discrimination. Journal of Computational Linguistics, 24:97–123, 1998.
  17. Fabian Suchanek, Gjergji Kasneci, and Gerhard Weikum. Yago - a large ontology from wikipedia and wordnet. Elsevier Journal of Web Semantics, 6(3):203–217, September 2008. find
  18. Antonio Toral, Rafael Muñoz, and Monica Monachini. Named entity wordnet. In Proceedings of the Sixth International Language Resources and Evaluation (LREC’08), Marrakech, Morocco, may 2008. find
  19. Dominic Widdows and Kathleen Ferraro. Semantic vectors: a scalable open source package and online technology management application. In Proceedings of the Sixth International Language Resources and Evaluation (LREC’08), Marrakech, Morocco, may 2008. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/toral-2009-study-linking-wikipedia-categories-wordnet/