Elamur Research⌕
☰
home / papers / 2012 / A Pipeline Arabic Named Entity Recognition using a Hybrid Approach

A Pipeline Arabic Named Entity Recognition using a Hybrid Approach

Paper · 2012 · Mai Oudah and Khaled Shaalan

Summary

Arabic NER is difficult because the script has no capitalization, words are agglutinative, short vowels are usually omitted, spellings vary, and annotated resources are scarce. Earlier Arabic systems used either handcrafted rules or machine learning. Rules are costly to maintain, and ML needs large tagged data. The authors combine the two in a pipeline. A rule-based component, a reworked and extended version of NERA, runs first. Its decisions then become input features for an ML classifier, along with morphological and contextual features. The system covers 11 entity types in three groups:

  • names (Person, Location, Organization),
  • temporal and numeric expressions (Date, Time, Price, Measurement, Percent),
  • technical strings (Phone Number, ISBN, File Name).

They test three classifiers (J48 decision trees, Libsvm, Logistic regression) on ACE2003/2004/2005, ATB Part1 v 2.0, ANERcorp and an in-house corpus. They also run ablations that drop feature groups. The hybrid beats both pure approaches and sets a new best on ANERcorp. This shows that symbolic knowledge and statistical learning work better together than either alone in a low-resource setting.

Key points

  • 11 NE types, split into 3 groups. Each group is evaluated on its own datasets: ACE2003/2004 (NW, BN) and ANERcorp for names; ACE2003–2005 and ATB Part1 v 2.0 for temporal and numeric types; the authors' own corpus for Phone Number, ISBN and File Name.
  • Best results on ANERcorp come from the hybrid with J48: F 0.944 for Person, 0.901 for Location and 0.882 for Organization. The previous best was the hybrid of Abdallah et al. (2012), with 0.928 / 0.874 / 0.861. ANERsys 1.0/2.0 and a CRF system score lower.
  • Removing the rule-based features ("W/O RB") hurts badly. For example, J48 on ACE2003 NW temporal and numeric types falls from about 0.984 to 0.735.
  • For temporal and numeric expressions, the hybrid is only slightly above the rule-based baseline. For the third group, rules alone already reach F = 1.0, and the hybrid matches that.
  • Decision trees (J48) were the most reliable classifier overall. Gazetteers and annotated corpora had to be collected by the authors.

Interesting ideas

  • I1. The rule system's output is not used as a final answer but as one feature among many, so the learner decides when to trust the rules.
  • I2. Ablation by feature family (all features / without rule features / without morphology) shows which knowledge source carries the result.
  • I3. Entity types are grouped by nature, and regular, pattern-like types (ISBN, phone numbers) are left largely to rules, where the rules are already perfect.
  • I4. The same surface form can be a common word or a name (e.g., "loyalty" vs. a person's name), and only context separates them.
  • I5. The authors plan to improve the rules by analyzing the hybrid system's output.

For Mova

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

  • I1 (tokens): Hand-written tagger or lemma heuristics could become features of the statistical tagger, so errors in the rules get corrected rather than passed down the pipeline.
  • I2 (learn): Evaluation could routinely include "minus one source" runs as controls, to prove that each component, such as rules or VerbNet roles, actually contributes.
  • I3 (world): Measures, values, dates and identifiers in a text's world could come from deterministic grammars, saving learned models for ambiguous names.
  • I4 (frames): Frame and coreference steps should be allowed to revise entity decisions using role context, instead of treating NER labels as fixed.
  • I5 (learn): Disagreements between the learned model and the rules could seed self-induced repair rules, each one explainable and checkable.

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 EdinburghEdinburgh · United Kingdom359 works · 77,352 citations🎓British University in DubaiDubai · United Arab Emirates10 works · 587 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 Mai Oudah and Khaled Shaalan. 2012. A Pipeline Arabic Named Entity Recognition using a Hybrid Approach. Proceedings of COLING 2012.

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

@inproceedings{oudah2012pipeline,
  title = {{A Pipeline Arabic Named Entity Recognition using a Hybrid Approach}},
  author = {Mai Oudah and Khaled Shaalan},
  booktitle = {Proceedings of COLING 2012},
  year = {2012},
  url = {https://aclanthology.org/C12-1132/}
}

References

34 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Abdallah, S., Shaalan, K. and Shoaib, M. (2012). Integrating Rule-based System with Classification for Arabic Named Entity Recognition. Proceedings of the 13th International Conference on Intelligent Text Processing and Computational Linguistics (CICLing), Springer-Verlag, Berlin Heidelberg, pages 311-322. find
  2. AbdelRahman, S., Elarnaoty, M., Magdy, M. and Fahmy, A. (2010). Integrated Machine Learning Techniques for Arabic Named Entity Recognition. International Journal of Computer Science Issues (IJCSI), Vol. 7, Issue 4, No 3, pages 27-36. find
  3. Abdul-Hamid, A. and Darwish, K. (2010). Simplified Feature Set for Arabic Named Entity Recognition. Proceedings of the 2010 Named Entities Workshop, (ACL 2010), pages 110-115.
  4. Babych, B. and Hartley, A. (2003). Improving Machine Translation Quality with Automatic Named Entity Recognition. Proceedings of the 7th International EAMT workshop on MT and other Language Technology Tools, Improving MT through other Language Technology Tools: Resources and Tools for Building MT (EAMT 2003), pages 1-8. find
  5. Benajiba, Y., Rosso, P. and Bened´i, J. M. (2007). ANERsys: An Arabic Named Entity Recognition system based on Maximum Entropy. Proceedings of the 8 th International Conference on Computational Linguistics and Intelligent Text Processing (CICLing-2007), Springer-Verlag, Berlin, Heidelberg, pages 143-153. find
  6. Benajiba, Y. and Rosso, P. (2007). ANERsys 2.0: Conquering the NER task for the Arabic language by combining the Maximum Entropy with POS-tag information. Proceedings of Workshop on Natural Language-Independent Engineering, 3rd Indian International Conference on Artificial Intelligence (IICAI-2007), pages 1814-1823. find
  7. Benajiba, Y. and Rosso, P. (2008). Arabic Named Entity Recognition using Conditional Random Fields. Proceedings of Workshop on HLT & NLP within the Arabic World (LREC 2008). find
  8. Benajiba, Y., Diab, M. and Rosso, P. (2008a). Arabic Named Entity Recognition: An SVM-Based Approach. Proceedings of Arab International Conference on Information Technology (ACIT 2008), pages 16-18. ↗
  9. Benajiba, Y., Diab, M. and Rosso, P. (2008b). Arabic Named Entity Recognition Using Optimized Feature Sets. Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP 2008), pages 284-293.
  10. Benajiba, Y., Diab, M. and Rosso, P. (2009a). Arabic Named Entity Recognition: A Feature-Driven Study. IEEE Transactions On Audio, Speech, And Language Processing, 17(5), pages 926-934. ↗
  11. Benajiba, Y., Diab, M. and Rosso, P. (2009b). Using Language Independent and Language Specific Features to Enhance Arabic Named Entity Recognition. The International Arab Journal of Information Technology, 6(5), pages 464- 473. find
  12. Cowie, J. and Wilks, Y. (1996). Information Extraction. Communications of the ACM, 39(1), pages 80-91. find
  13. Elsebai, A., Meziane, F. and BelKredim, F. Z. (2009). A Rule Based Persons Names Arabic Extraction System. Communications of the IBIMA, pages 53-59. find
  14. Farber, B., Freitag, D., Habash, N. and Rambow, O. (2008). Improving NER in Arabic Using a Morphological Tagger. Proceedings of Workshop on HLT & NLP within the Arabic World (LREC 2008), pages 2509- 2514.
  15. Habash, N. and Rambow, O. (2005). Arabic Tokenization, Part-of-Speech Tagging and Morphological Disambiguation in One Fell Swoop. Proceedings of the 43rd Annual Meeting of the Association for Computational Linguistics (ACL'05), pages 573-580.
  16. Habash, N., Soudi, A. and Buckwalter, T. (2007). On Arabic Transliteration. Arabic Computational Morphology: Knowledge-based and Empirical Methods, Springer, pages 15-22. ↗
  17. Hamadene, A., Shaheen, M. and Badawy, O. (2011). ARQA: An Intelligent Arabic Question Answering System. Proceedings of Arabic Language Technology International Conference (ALTIC 2011). find
  18. Hastie, T., Tibshirani, R. and Friedman, J. (2009). The Elements of Statistical Learning: Data Mining, Inference, and Prediction. (2 nd ed.). Springer. find
  19. Maamouri, M., Bies, A., Jin, H. and Buckwalter, T. (2003). Arabic Treebank: Part 1 v 2.0. LDC2003T06: Linguistic Data Consortium, Philadelphia. ↗
  20. Mesfar, S. (2007). Named Entity Recognition for Arabic Using Syntactic Grammars. Proceedings of the 12 th International Conference on Application of Natural Language to Information Systems, Springer-Verlag, Berlin, Heidelberg, pages 305-316. find
  21. Mitchell, A., Strassel, S., Huang, S., and Zakhary, R. (2005). ACE 2004 Multilingual Training Corpus. Ldc2005t09: Linguistic Data Consortium, Philadelphia. find
  22. Mitchell, A., Strassel, S., Przybocki, M., Davis, J., Doddington, G., Grishman, R., Meyers, A., Brunstein, A., Ferro, L. and Sundheim, B. (2003). Tides Extraction (ACE) 2003 Multilingual Training Data. Ldc2004t09: Linguistic Data Consortium, Philadelphia. ↗
  23. Molla, D., Zaanen, M. and Smith, D. (2006). Named Entity Recognition for Question Answering. In Proceedings of the 2006 Australasian Language Technology Workshop (ALTW2006), pages 51-58. find
  24. Nadeau, D. and Sekine, S. (2007). A Survey of Named Entity Recognition and Classification. Lingvisticae Investigationes, 30(1), pages 3-26. ↗
  25. Orphanos, G., Kalles, D., Papagelis, T. and Christodoulakis, D. (1999). Decision Trees and NLP: A Case Study in POS Tagging. Proceedings of Annual Conference on Artificial Intelligence (ACAI). find
  26. Petasis, G., Vichot, F., Wolinski, F., Paliouras, G., Karkaletsis, V. and Spyropoulos, C. D. (2001). Using Machine Learning to Maintain Rule-based Named-Entity Recognition and Classification Systems. Proceeding Conference of Association for Computational Linguistics, pages 426-433. find
  27. Roth, R., Rambow, O., Habash, N., Diab, M. and Rudin, C. (2008). Arabic Morphological Tagging, Diacritization, and Lemmatization Using Lexeme Models and Feature Ranking. Proceedings of the 46 th Annual Meeting of the Association for Computational Linguistics on Human Language Technologies: Short Papers (HLT-Short '08), pages 117-120.
  28. Shaalan, K. (2010). Rule-based Approach in Arabic Natural Language Processing. The International Journal on Information and Communication Technologies (IJICT), 3(3), pages 11-19. find
  29. Shaalan, K. and Raza, H. (2007). Person Name Entity Recognition for Arabic. Proceedings of the 5th Workshop on Important Unresolved Matters, pages 17-24.
  30. Shaalan, K. and Raza, H. (2008). Arabic Named Entity Recognition from Diverse Text Types. Proceedings of the 6th International Conference on Natural Language Processing (GoTAL 2008), Springer-Verlag, Berlin, Heidel-berg, pages 440-451. find
  31. Shaalan, K. and Raza, H. (2009). NERA: Named Entity Recognition for Arabic. Journal of the American Society for Information Science and Technology, 60(8), pages 1652–1663. ↗
  32. Vapnik, V. N. (1995). The Nature of Statistical Learning Theory. Springer-Verlag New York, Inc. ↗
  33. Walker, C., Strassel, S., Medero, J., and Maeda, K. (2006). ACE 2005 Multilingual Training Corpus. Ldc2006t06: Linguistic Data Consortium, Philadelphia. find
  34. Zaghouani, W. (2012). RENAR: A Rule-Based Arabic Named Entity Recognition System. ACM Transactions on Asian Language Information Processing, 11(1), Article no. 2, pages 1-13. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/oudah-2012-pipeline-arabic-named-entity-recognition/