Elamur Research⌕
☰
home / papers / 2014 / Named Entity Recognition System for Dialectal Arabic

Named Entity Recognition System for Dialectal Arabic

Paper · 2014 · Ayah Zirikly and Mona Diab

Summary

Most Arabic NER work targets Modern Standard Arabic (MSA) newswire. Those systems transfer poorly to Dialectal Arabic (DA), which has no standard spelling, little annotated data, and weak gazetteer coverage. The authors build an NER system for the Egyptian dialect (EGY) and release a manually annotated EGY dataset, available from the authors on request.

The system is a supervised CRF sequence labeler. It combines:

  • character n-gram and context features,
  • dialect trigger words near a name,
  • morphological features from MADAMIRA,
  • several gazetteer-matching schemes, including fuzzy matching by Levenshtein distance,
  • optionally, Brown clusters.

Two baselines are reimplemented: BAS1 (Abdul-Hamid and Darwish, 2010) and BAS2 (Shaalan and Oudah, 2014), the latter without its rule component. The best feature set reaches a weighted macro F1 of about 70.3%, against 54.8% for BAS1. This is the roughly 16-point gain claimed. LOC is recognized far more reliably than PER. The work shows that features chosen for dialect-specific traits, especially spelling-tolerant lexicon lookup, clearly beat MSA-oriented feature sets on informal Arabic.

Key points

  • Baselines: BAS1 reaches F1 54.762% (LOC 76.191, PER 33.333). BAS2 manages only 31%, despite being state of the art on MSA ANERcorp.
  • Best configuration: FEA6 (all features except Brown clustering) gives overall F1 70.305%, with LOC 91.429 and PER 49.18.
  • Feature gains:
    • Trigger keywords at distance 1 and 2 (KEY) add about 2 points absolute. The pattern Abw+name covers about 46% of PER names in the dialect.
    • Morphological features (GEN, POS, CAPS) add about 9 points.
  • Brown clustering: adding BC drops F1 to 65.758. Same-class locations such as mSr, AmrykA and qtr received cluster bit strings with almost no shared prefix.
  • Gazetteers: they use ANERGaz among others, matched exactly (EM-GAZ), partially (PM-GAZ) and by Levenshtein distance (LVM-GAZ). LVM-GAZ helps mainly PER, because family names vary in spelling (e.g. with or without the Al prefix).

Interesting ideas

  • I1. Lexicon lookup with edit-distance tolerance absorbs unstandardized spelling variants of names, so a fixed gazetteer still works on messy text.
  • I2. A capitalization cue is recovered for a script that lacks case by checking whether the analyzer's English gloss of a word is capitalized, borrowing a signal from another language's resource.
  • I3. Trigger words within one or two tokens of a candidate (like the kinship prefix Abw before a name) are cheap, explainable evidence for the PER class.
  • I4. Before trusting a distributional feature, check whether known same-class items actually share structure in it; here shared Brown-cluster prefixes exposed why BC hurt.
  • I5. Per-class error analysis links the PER/LOC gap to measurable data properties, namely the singleton rate and how consistently a string keeps its tag (mSr always LOC; EAdl both an adjective and a name).

For Mova

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

  • I1 (world): when linking entity mentions to Wikidata/WordNet entries, fall back to a bounded Levenshtein match so misspelled or variant character names in tales and tests still resolve, while logging the distance as part of the explanation.
  • I2 (tokens): when English casing is unreliable (all-caps headlines, sentence-initial words), consult the WordNet/lexicon entry to decide whether a token is normally a proper noun instead of trusting surface case.
  • I3 (frames): titles and kinship words such as "Mr", "King" or "Uncle" can be read off the UD tree as modifiers that mark the head as a person, feeding role assignment and coreference.
  • I4 (learn): add a sanity test that verifies same-class seed words cluster together before any embedding or cluster feature enters training, acting as a negative control.
  • I5 (learn): report the singleton ratio and label-consistency statistics next to per-class scores in evaluations, so weak categories are explained by data properties rather than guessed at.

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

 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 Ayah Zirikly and Mona Diab. 2014. Named Entity Recognition System for Dialectal Arabic. Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP).

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.3115/v1/w14-3610

@inproceedings{zirikly2014named,
  title = {{Named Entity Recognition System for Dialectal Arabic}},
  author = {Ayah Zirikly and Mona Diab},
  booktitle = {Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP)},
  year = {2014},
  doi = {10.3115/v1/w14-3610},
  url = {https://doi.org/10.3115/v1/w14-3610}
}

References

25 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Sherief Abdallah, Khaled Shaalan, and Muhammad Shoaib. 2012. Integrating rule-based system with classification for arabic named entity recognition. In Computational Linguistics and Intelligent Text Processing, pages 311–322. Springer. find
  2. Ahmed Abdul-Hamid and Kareem Darwish. 2010. Simplified feature set for arabic named entity recognition. In Proceedings of the 2010 Named Entities Workshop, NEWS ’10, pages 110–115, Stroudsburg, PA, USA. Association for Computational Linguistics.
  3. Yassine Benajiba and Paolo Rosso. 2008. Arabic named entity recognition using conditional random fields. In Proc. of Workshop on HLT & NLP within the Arabic World, LREC, volume 8, pages 143–153. Citeseer. find
  4. Yassine Benajiba, Paolo Rosso, and José-Miguel Benedı́. 2007. Anersys: An arabic named entity recognition system based on maximum entropy. In CICLing, pages 143–153. find
  5. Yassine Benajiba, Mona Diab, and Paolo Rosso. 2008. Arabic named entity recognition using optimized feature sets. In Proceedings of the Conference on Empirical Methods in Natural Language Processing, pages 284–293. Association for Computational Linguistics.
  6. Yassine Benajiba, Mona Diab, and Paolo Rosso. 2009. Arabic named entity recognition: A feature-driven study. Audio, Speech, and Language Processing, IEEE Transactions on, 17(5):926–934. ↗
  7. Yassine Benajiba, Imed Zitouni, Mona Diab, and Paolo Rosso. 2010. Arabic named entity recognition: Using features extracted from noisy data. In Proceedings of the ACL 2010 Conference Short Papers, ACLShort ’10, pages 281–285, Stroudsburg, PA, USA. Association for Computational Linguistics.
  8. Peter F Brown, Peter V Desouza, Robert L Mercer, Vincent J Della Pietra, and Jenifer C Lai. 1992. Class-based n-gram models of natural language. Computational linguistics, 18(4):467–479. find
  9. Kareem Darwish and Wei Gao. 2014. Simple effective microblog named entity recognition: Arabic as an example. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC-2014), Reykjavik, Iceland, May 26-31, 2014., pages 2513–2517.
  10. Ali Elsebai, Farid Meziane, and Fatma Zohra Belkredim. 2009. A rule based persons names arabic extraction system. Communications of the IBIMA, 11(6):53–59. find
  11. Sergio Ferrndez, Antonio Toral, scar Ferrndez, Antonio Ferrndez, and Rafael Muoz. 2007. Applying wikipedias multilingual knowledge to cross-lingual question answering. In In Zoubida Kedad, Nadira Lammari, Elisabeth Mtais, Farid Meziane, and Yacine Rezgui, editors, NLDB, volume 4592 of Lecture Notes in Computer Science. Springer. find
  12. Nizar Habash, Mona T Diab, and Owen Rambow. 2012. Conventional orthography for dialectal arabic. In LREC, pages 711–718.
  13. Nizar Habash, Ryan Roth, Owen Rambow, Ramy Eskander, and Nadi Tomeh. 2013. Morphological analysis and disambiguation for dialectal arabic. In Human Language Technologies: Conference of the North American Chapter of the Association of Computational Linguistics, Proceedings, June 9-14, 2013, Westin Peachtree Plaza Hotel, Atlanta, Georgia, USA, pages 426–432.
  14. John Lafferty, Andrew McCallum, and Fernando CN Pereira. 2001. Conditional random fields: Probabilistic models for segmenting and labeling sequence data. find
  15. Vladimir I Levenshtein. 1966. Binary codes capable of correcting deletions, insertions and reversals. In Soviet physics doklady, volume 10, page 707. find
  16. David Nadeau and Satoshi Sekine. 2007. A survey of named entity recognition and classification. Lingvisticae Investigationes, 30(1):3–26. ↗
  17. David Nadeau, Peter Turney, and Stan Matwin. 2006. Unsupervised named-entity recognition: Generating gazetteers and resolving ambiguity. find
  18. Naoaki Okazaki. 2007. Crfsuite: A fast implementation of conditional random fields (crfs). find
  19. Arfath Pasha, Mohamed Al-Badrashiny, Ahmed El Kholy, Ramy Eskander, Mona Diab, Nizar Habash, Manoj Pooleery, Owen Rambow, and Ryan Roth. 2014. Madamira: A fast, comprehensive tool for morphological analysis and disambiguation of arabic. In In Proceedings of the 9th International Conference on Language Resources and Evaluation, Reykjavik, Iceland.
  20. Karin C Ryding. 2005. A Reference Grammar of Modern Standard Arabic. Cambridge University Press. ↗
  21. Khaled Shaalan and Mai Oudah. 2014. A hybrid approach to arabic named entity recognition. Journal of Information Science, 40(1):67–87. ↗
  22. Khaled Shaalan and Hafsa Raza. 2009. Nera: Named entity recognition for arabic. Journal of the American Society for Information Science and Technology, 60(8):1652–1663. ↗
  23. Khaled Shaalan. 2014. A survey of arabic named entity recognition and classification. Comput. Linguist., 40(2):469–510, June. ↗
  24. Paul Thompson and Christopher C. Dozier. 1997. Name searching and information retrieval. In In Proceedings of Second Conference on Empirical Methods in Natural Language Processing, pages 134–140. ↗
  25. Joseph Turian, Lev Ratinov, and Yoshua Bengio. 2010. Word representations: A simple and general method for semi-supervised learning. In Proceedings of the 48th Annual Meeting of the Association for Computational Linguistics, pages 384–394. Association for Computational Linguistics.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zirikly-2014-named-entity-recognition-system-dialectal/