Elamur Research⌕
☰
home / papers / 2015 / Named Entity Recognition of Persons' Names in Arabic Tweets

Named Entity Recognition of Persons' Names in Arabic Tweets

Paper · 2015 · Omnia Zayed and Samhaa El-Beltagy

Summary

The paper addresses recognising person names in Arabic tweets. These are mostly written in colloquial dialects, where tools built for Modern Standard Arabic (MSA) perform poorly. Arabic has no capital letters and usually omits diacritics, and clitics attach freely to words. As a result, many names share their surface form with ordinary words, verbs, prepositions or dialect expressions.

The authors avoid parsers, taggers and morphological analysers entirely. They combine a rule-based component with a statistical one. The statistical part learns scored unigram context patterns without supervision, using association rules that signal a nearby name. Dictionaries of names, built from a list of full names taken from public sources such as Wikipedia, are grouped into clusters. These clusters act as constraints on which name parts may combine.

The system was tested on Darwish's tweet test set with the strict CoNLL scorer. It beat earlier systems, with a relative F-score gain of 19.6% over Darwish and Gao (2014). This matters because the method needs few resources and should transfer to other languages, genres and domains.

Key points

  • Test data: Darwish's test set, 1,423 tweets with roughly 26k tokens, covering Egyptian, Levantine and Gulf dialects, with both Arabic and English entities tagged.
  • Scoring used the CoNLL conlleval script with exact-match credit only (no reward for partially found names), on a single class: persons' names.
  • On the unmodified data (E1), the system reached P 81.93%, R 56.32%, F 66.75%. The comparison systems scored 59.59% (Zayed and El-Beltagy 2015), 55.8% (Darwish and Gao 2014) and 39.8% (Darwish 2013).
  • Four setups were tested: annotation errors fixed or not, and English entities removed or not. The best (E4) gave P 85.40%, R 65.17%, F 73.92%.
  • Correcting annotation mistakes added about 4.85% to the score, and excluding English entities raised recall by 5.89%.

Interesting ideas

  • I1. Instead of trusting dictionary hits, a candidate name is accepted only when backed by context patterns that were scored and learned without supervision, which filters out ordinary words that look like names.
  • I2. Clustering name lists and allowing a multi-word name only when its parts fall in the same cluster turns a flat lexicon into a cheap combinatory constraint against spurious spans.
  • I3. The authors deliberately avoid parsers and morphology so the method survives noisy, dialectal text where those tools break, using only a name list from public sources.
  • I4. The evaluation reports several variants (raw data, fixed gold errors, out-of-scope entities removed) with a strict exact-match scorer, separating system errors from dataset noise and task-scope mismatch.

For Mova

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

  • I1 (world): Entity creation could require a scored contextual cue (title, verb of address, appositive) before promoting a capitalised or dictionary token to a person, with the cue logged as the explanation.
  • I2 (tokens): Multiword proper-noun grouping could check that given-name and surname parts come from compatible clusters (e.g., the same origin), rejecting accidental bigrams like "Will Power" in ordinary prose.
  • I3 (learn): A lightweight, parser-independent name detector could act as an extra witness when building silver data, flagging tweets or informal text where the parser's PROPN decisions disagree with it.
  • I4 (learn): Mova's evaluations could routinely report raw-gold, corrected-gold and in-scope-only scores, so improvements are not confused with annotation artefacts.

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

🎓Nile UniversityGiza · Egypt14 works · 158 citations

Also on the byline: IEEE Computer Society.

 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 Omnia Zayed and Samhaa El-Beltagy. 2015. Named Entity Recognition of Persons' Names in Arabic Tweets. Proceedings of the International Conference Recent Advances in Natural Language Processing.

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

@inproceedings{zayed2015named,
  title = {{Named Entity Recognition of Persons' Names in Arabic Tweets}},
  author = {Omnia Zayed and Samhaa El-Beltagy},
  booktitle = {Proceedings of the International Conference Recent Advances in Natural Language Processing},
  year = {2015},
  url = {https://aclanthology.org/R15-1093/}
}

References

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

  1. Rakesh Agrawal, Tomasz Imieliński, and Arun Swami. 1993. Mining association rules between sets of items in large databases. In Proceedings of the 1993 ACM SIGMOD international conference on Management of data, SIGMOD 1993, number May, pages 207–216, New York, USA. ACM Press. ↗
  2. Yassine Benajiba, Mona Diab, and Paolo Rosso. 2009. Arabic Named Entity Recognition: A Feature-Driven Study. IEEE Transactions on Audio, Speech, and Language Processing, 17(5):926–934, July. ↗
  3. Vincent D Blondel, Jean-Loup Guillaume, Renaud Lambiotte, and Etienne Lefebvre. 2008. Fast unfolding of communities in large networks. Journal of Statistical Mechanics: Theory and Experiment, 2008(10):P10008, October. ↗
  4. Kareem Darwish. 2013. Named Entity Recognition using Cross-lingual Resources: Arabic as an Example. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics, volume 1, pages 1558–1567, Sofia, Bulgaria. Association for Computational Linguistics.
  5. 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), pages 2513–2517, Reykjavik, Iceland, May. European Language Resources Association (ELRA).
  6. A De Sitter, T Calders, and Walter Daelemans. 2004. A formal framework for evaluation of information extraction. Technical report, Antwerp. find
  7. A Farghaly and K Shaalan. 2009. Arabic natural language processing: Challenges and solutions. ACM Transactions on Asian Language Information Processing (TALIP), 8(4):1–22. ↗
  8. Nizar Y. Habash. 2010. Introduction to Arabic Natural Language Processing. Mogran & Claypool. ↗
  9. Daniel Jurafsky and James H. Martin. 2009. Information Extraction. In Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics, and Speech Recognition., chapter 22, pages 725–743. Prentice Hall, 2nd edition. find
  10. Alan Ritter, Sam Clark, and Oren Etzioni. 2011. Named Entity Recognition in Tweets : An Experimental Study. In Conference on Empirical Methods in Natural Language Processing, pages 1524–1534.
  11. Khaled Shaalan. 2014. A survey of arabic named entity recognition and classification. Computational Linguistics, 40(2):469–510, June. ↗
  12. Carolin Shihadeh and Günter Neumann. 2012. ARNE: A tool for namend entity recognition from Arabic text. In Fourth Workshop on Computational Approaches to Arabic Script-based Languages (CAASL4), San Diego, CA, USA.
  13. Amit Singhal. 2001. Modern Information Retrieval: A Brief Overview. Bulletin of the IEEE Computer Society Technical Committee on Data Engineering, 24(4):35–43. find
  14. Omnia Zayed, Samhaa El-Beltagy, and Osama Haggag. 2013. An approach for extracting and disambiguating arabic persons‟ names using clustered dictionaries and scored patterns. In Elisabeth Métais, Farid Meziane, Mohamad Saraee, Vijayan Sugumaran, and Sunil Vadera, editors, NLDB 2013, LNCS, volume 7934, pages 201–212. Springer Berlin Heidelberg. find
  15. Omnia H. Zayed and Samhaa R. El-Beltagy. 2015. A Hybrid Approach for Extracting Arabic Persons‟ Names and Resolving their Ambiguity from Twitter. In 20th International Conference on Application of Natural Language to Information Systems (NLDB 2015), Passau, Germany, June. Springer. find
  16. Ayah Zirikly and Mona Diab. 2014. Named Entity Recognition System for Dialectal Arabic. In Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP), pages 78– 86, Doha, Qatar, October. Association for Computational Linguistics.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zayed-2015-named-entity-recognition-persons-names/