Elamur Research⌕
☰
home / papers / 2013 / Automatically Developing a Fine-grained Arabic Named Entity Corpus and Gazetteer by utilizing Wikipedia

Automatically Developing a Fine-grained Arabic Named Entity Corpus and Gazetteer by utilizing Wikipedia

Paper · 2013 · Fahd Alotaibi and Mark Lee

Summary

Arabic named entity recognition has mostly used a few coarse classes, and fine-grained annotated data is expensive to create by hand. The authors build such data from Arabic Wikipedia with very little human effort. They adapt the two-level ACE (2008) taxonomy to an open-domain encyclopedia by adding a Product class and occupation-based subtypes for persons. They hand-label 4000 articles and train SVM and SGD classifiers that assign each article a coarse and a fine NE tag. These article labels are then carried over to the hyperlinked phrases that point to each article. Later mentions without links are also found, with Arabic morphological variation taken into account. Finally, suitable sentences are selected. The outputs are two corpus versions, WikiFANE Whole and WikiFANE Selective, plus a gazetteer, WikiFANE Gazet. When the gazetteer is added to a CRF tagger, scores rise by at least 2 F-points on news-wire text, and the best within-Wikipedia result is 85.18% F. The work shows that the link structure of an encyclopedia can stand in for costly manual NE annotation.

Key points

  • The tagset has eight coarse classes: PER, ORG, LOC, GPE, FAC, VEH, WEA, and the new PRO. Fine subtypes include Politician, Athlete, Software, and Drug. Because there are two levels, it can be collapsed to CoNLL- or MUC-style schemes.
  • Two native speakers annotated 4000 articles. Kappa agreement improved as the sample grew, reaching 99 (coarse) and 97 (fine) at n=4000.
  • Article features build on Alotaibi and Lee (2012): SF, FF, LF (stems), and ELF (POS-filtered). Two features are new: text from the whole first paragraph and bigrams. All features are weighted with tf-idf, and the data is split 80/20.
  • In the within-domain CRF tests, WikiFANE Gazet adds at least 3 F-points. WikiFANE Selective with the gazetteer scores P 88.77, R 81.86, and F 85.18. LOC is the weakest class.
  • When the tagger is tested across domains, recall falls sharply while precision stays high.

Interesting ideas

  • I1. Instead of tagging every running word, classify each target page once and let every hyperlink pointing to that page inherit its label, so one decision spreads to many mentions in context.
  • I2. A two-level tagset whose fine classes collapse cleanly into coarse ones lets a single resource serve several older annotation schemes and stay comparable with them.
  • I3. Agreement was measured at several sample sizes (500, 2000, 4000), which shows how quickly annotators settle and where the guidelines are unclear early on.
  • I4. After linked mentions are labeled, later unlinked mentions of the same entity are also picked up, allowing for morphological variants of the name.
  • I5. The authors release both a full corpus and a filtered "selective" one, and they test the gazetteer by running each setup with and without it.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (learn): Wikipedia link anchors could supply silver entity spans for training, with each link resolved through its target page's Wikidata type.
  • I2 (world): the entity ontology could store fine types with a guaranteed upward mapping, so answers and evaluations can be reported at whichever granularity a test needs.
  • I3 (learn): when several witness parsers or LLM teacher tiers produce silver labels, tracking their agreement as the batch grows could indicate when a label category is stable enough to trust.
  • I4 (frames): coreference could use the same trick, propagating an entity's type from its first anchored mention to its later short or inflected forms.
  • I5 (learn): every silver dataset could ship in strict and loose versions, with the knowledge-source ablation kept as a routine control.

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 BirminghamBirmingham · United Kingdom25 works · 62,564 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

🧭 Topics 2Research subjects and languages the work is about.Semitic languages · 2,908 worksNamed entity recognition · 200 works

Cite

Cite as Fahd Alotaibi and Mark Lee. 2013. Automatically Developing a Fine-grained Arabic Named Entity Corpus and Gazetteer by utilizing Wikipedia. Proceedings of the Sixth International Joint Conference on Natural Language Processing.

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

@inproceedings{alotaibi2013automatically,
  title = {{Automatically Developing a Fine-grained Arabic Named Entity Corpus and Gazetteer by utilizing Wikipedia}},
  author = {Fahd Alotaibi and Mark Lee},
  booktitle = {Proceedings of the Sixth International Joint Conference on Natural Language Processing},
  year = {2013},
  url = {https://aclanthology.org/I13-1045/}
}

References

29 references, 5 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. ACE. 2008. Ace (automatic content extraction) english annotation guidelines for entities, 06. [accessed 10 April 2013]. find
  2. Fahd Alotaibi and Mark Lee. 2012. Mapping Arabic Wikipedia into the named entities taxonomy. In Proceedings of COLING 2012: Posters, pages 43– 52, Mumbai, India, December.
  3. The COLING 2012 Organizing Committee. find
  4. Joohui An, Seungwoo Lee, and Gary Geunbae Lee. 2003. Automatic acquisition of named entity tagged corpus from world wide web. In Proceedings of the 41st Annual Meeting on Association for Computational Linguistics, pages 165–168. Association for Computational Linguistics. find
  5. Mohammed Attia, Antonio Toral, Lamia Tounsi, Monica Monachini, and Josef van Genabith. 2010. An automatically built named entity lexicon for arabic. In Proceedings of the Seventh conference on International Language Resources and Evaluation (LREC10), Valletta, Malta. European Language Resources Association (ELRA).
  6. Yassine Benajiba, Paolo Rosso, and José Miguel Benedı́ruiz. 2007. Anersys: An arabic named entity recognition system based on maximum entropy. In Alexander Gelbukh, editor, Computational Linguistics and Intelligent Text Processing, volume 4394 of Lecture Notes in Computer Science, pages 143–153. Springer Berlin / Heidelberg. find
  7. Y. Benajiba, I. Zitouni, M. Diab, and P. Rosso. 2010. Arabic named entity recognition: Using features extracted from noisy data. In Proceedings of the ACL 2010 Conference Short Papers, pages 281–285, Uppsala, Sweden. Association for Computational Linguistics.
  8. Ada Brunstein. 2002. Annotation guidelines for answer types. LDC2005T33 [accessed 02 January 2012]. find
  9. Jean Carletta. 1996. Assessing agreement on classification tasks: the kappa statistic. Comput. Linguist., 22(2):249–254, June. ↗
  10. Xavier Carreras, Lluis Marquez, and Lluı́s Padró. 2002. Named entity extraction using adaboost. In proceedings of the 6th conference on Natural language learning-Volume 20, pages 1–4. Association for Computational Linguistics. find
  11. Wisam Dakka and Silviu Cucerzan. 2008. Augmenting wikipedia with named entity tags. In Proceedings of the 3rd International Joint Conference on Natural Language Processing, pages 545–552, Hyderabad, India. Asian Federation of Natural Language Processing. find
  12. Mona Diab. 2009. Second generation amira tools for arabic processing: Fast and robust tokenization, pos tagging, and base phrase chunking. In 2nd International Conference on Arabic Language Resources and Tools. find
  13. Maud Ehrmann, Marco Turchi, and Ralf Steinberger. 2011. Building a multilingual named entity-annotated corpus using annotation projection. In Proceedings of Recent Advances in Natural Language Processing (RANLP), pages 118–124, Hissar, Bulgaria. find
  14. Ruiji Fu, Bing Qin, and Ting Liu. 2011. Generating chinese named entity data from a parallel corpus. In Proceedings of the 5th International Joint Conference on Natural Language Processing, pages 264–272, Chiang Mai, Thailand. Asian Federation of Natural Language Processing (AFNLP). find
  15. Nizar Habash, Abdelhadi Soudi, and Tim Buckwalter. 2007. On arabic transliteration. In Arabic Computational Morphology, volume 38 of Text, Speech and Language Technology, pages 15–22. Springer Netherlands. ↗
  16. Sanda Harabagiu, Dan Moldovan, Christine Clark, Mitchell Bowden, John Williams, and Jeremy Bensley. 2003. Answer mining by combining extraction techniques with abductive reasoning. In Proceedings of 12th Text Retrieval Conference, volume 2003, pages 375–382. NIST. find
  17. Xin Li and Dan Roth. 2006. Learning question classifiers: the role of semantic information. Natural Language Engineering, 12(03):229–249. ↗
  18. David Nadeau, Peter D. Turney, and Stan Matwin. 2006. Unsupervised named-entity recognition: Generating gazetteers and resolving ambiguity. Advances in Artificial Intelligence, pages 266–277. find
  19. Elisa Noguera, Antonio Toral, Fernando Llopis, and Rafael Muńoz. 2005. Reducing question answering input data using named entity recognition. In Text, Speech and Dialogue, pages 428–434. Springer. find
  20. Joel Nothman, James Curran, and Tara Murphy. 2008. Transforming wikipedia into named entity training data. In Proceedings of the Australian Language Technology Association Workshop, pages 124–132, Hobart, Australia. ALTA. find
  21. Joel Nothman, Tara Murphy, and James Curran. 2009. Analysing wikipedia and gold-standard corpora for ner training. In Proceedings of the 12th Conference of the European Chapter of the Association for Computational Linguistics, pages 612–620, Athens, Greece. Association for Computational Linguistics. find
  22. Alexander Richman and Patrick Schon. 2008. Mining wiki resources for multilingual named entity recognition. In Proceedings of the 46th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, pages 1–9, Columbus, Ohio, USA. Association for Computational Linguistics. find
  23. Iman Saleh, Kareem Darwish, and Aly Fahmy. 2010. Classifying wikipedia articles into ne’s using svm’s with threshold adjustment. In Proceedings of the 2010 Named Entities Workshop, pages 85–92, Uppsala, Sweden. Association for Computational Linguistics. find
  24. Satoshi Sekine and Chikashi Nobat. 2004. Definition, dictionaries and tagger for extended named entity hierarchy. In Proceedings of the 4th International Conference on Language Resources And Evaluation, pages 1977–1980, Lisbon, Portugal. ELRA. find
  25. Satoshi Sekine, Kiyoshi Sudo, and Chikashi Nobata. 2002. Extended named entity hierarchy. In Proceedings of the third International Conference on Language Resources and Evaluation, volume 2, Las Palmas, Spain. ELRA. find
  26. Khaled Shaalan and Hafsa Raza. 2007. Person name entity recognition for arabic. In Proceedings of the 2007 Workshop on Computational Approaches to Semitic Languages: Common Issues and Resources, pages 17–24, Prague, Czech Republic. Association for Computational Linguistics.
  27. Sam Tardif, James Curran, and Tara Murphy. 2009. Improved text categorisation for wikipedia named entities. In Australasian Language Technology Association Workshop 2009, pages 104–108, Sydney, Australia. find
  28. Maksim Tkatchenko, Alexander Ulanov, and Andrey Simanovsky. 2011. Classifying wikipedia entities into fine-grained classes. In Data Engineering Workshops (ICDEW), 2011 IEEE 27th International Conference on, pages 212–217. IEEE. find
  29. Wikipedia. 2013. The statistic of arabic wikipedia, 05. [accessed 10 May 2013]. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/alotaibi-2013-automatically-developing-fine-grained-arabic/