---
title: "Improving NER in Arabic Using a Morphological Tagger (2008)"
description: "The authors test whether a full morphological tagger can make up for these gaps. 79.8% F1. 13 references, cited by 15 papers here."
authors: ["Benjamin Farber", "Dayne Freitag", "Nizar Habash", "Owen Rambow"]
year: 2008
venue: "Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC'08)"
source: "https://aclanthology.org/L08-1054/"
licence: "CC BY-NC 4.0 (ELRA, LREC proceedings)"
licence_url: "https://creativecommons.org/licenses/by-nc/4.0/"
summary_licence: "Elamur Cosmic License"
canonical: "https://research.elamur.ai/free/en/papers/farber-2008-improving-ner-arabic-morphological-tagger/"
---
[Elamur Research](https://md.research.elamur.ai/free/en/) · [this page in HTML](https://research.elamur.ai/free/en/papers/farber-2008-improving-ner-arabic-morphological-tagger/)

[home](https://md.research.elamur.ai/free/en/) / [papers](https://md.research.elamur.ai/free/en/papers/) / [2008](https://md.research.elamur.ai/free/en/papers/2008/) / Improving NER in Arabic Using a Morphological Tagger

# Improving NER in Arabic Using a Morphological Tagger

**Authors:** Benjamin Farber, Dayne Freitag, [Nizar Habash](https://md.research.elamur.ai/free/en/authors/nizar-habash/) and [Owen Rambow](https://md.research.elamur.ai/free/en/authors/owen-rambow/) · **Year:** 2008 · **Venue:** [Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC'08)](https://md.research.elamur.ai/free/en/venues/lrec/) ([LREC 2008](https://md.research.elamur.ai/free/en/venues/lrec/2008/)) · **Publisher:** [European Language Resources Association (ELRA)](https://md.research.elamur.ai/free/en/publishers/european-language-resources-association-elra/) · **Licence:** [CC BY-NC 4.0 (ELRA, LREC proceedings)](https://creativecommons.org/licenses/by-nc/4.0/)

**Links:** [Publisher's page](https://aclanthology.org/L08-1054/) · [PDF](https://aclanthology.org/L08-1054.pdf) · [OpenAlex](https://openalex.org/W68059783)

**Citations:** 43 citations (OpenAlex, 2026-10-14) · [15](https://md.research.elamur.ai/free/en/papers/farber-2008-improving-ner-arabic-morphological-tagger/cited-by/) in our library, of the 149,439 papers whose bibliographies we read

## Summary
Arabic named entity recognition is harder than English: annotated data is scarcer, written words are highly ambiguous because short vowels are omitted and clitics are attached, and there is no capitalization to signal names. The authors test whether a full morphological tagger can make up for these gaps. Their baseline is a largely language-neutral structured perceptron (Collins, 2002) with BIO labels. It uses word-identity features in a ±3 token window and distributional word clusters built from Arabic Gigaword, and it is trained on ACE 2005 Arabic newswire. They add two feature types derived from BAMA analyses or from MADA's disambiguated choice. One fires when the English gloss of the analysis is capitalized; the other fires when the word is out of vocabulary for the analyzer. With MADA, F1 rises from 0.667 to 0.715, roughly a 14% error reduction, and recall gains most. Error analysis and simple system combinations show how to shift the balance between precision and recall. The work shows that lexical disambiguation from a morphological tagger is a useful substitute for capitalization when finding names.

## Key points
- Data: newswire part of the ACE 2005 Arabic training set, restricted to name mentions (mostly persons, organizations, GPEs). Tokenization is deliberately naive: letter runs, digit runs, and single punctuation marks.
- Tools: BAMA (Buckwalter analyzer, over 35K lexemes) proposes all analyses; the word *byn* alone has 19. MADA, trained on PATB part 3, picks one analysis with over 96% reported accuracy.
- Results (F1/P/R): Base 0.667/0.714/0.627; BAMA 0.676/0.713/0.643; MADA 0.715/0.735/0.697.
- Errors are mostly missed entities, often common GPEs. In nearly half of span errors the predicted span is shorter than the gold one, and 82% of label errors involve GPEs.
- The systems agree with each other at only 76.5–81.7% F1. An oracle choosing the correct system reaches 79.8% F1. Union yields 73.0% recall and Intersection yields 84.3% precision, but neither beats MADA on F1.

## Interesting ideas
- Whether the tagger's selected analysis has an English gloss written with a capital letter serves as a stand-in signal that the word is a name, borrowing an orthographic cue from another language's lexicon.
- Whether the morphological analyzer recognizes a word at all becomes a feature, since unknown words are disproportionately names.
- The study compares the full set of analyses (BAMA) against the single disambiguated analysis (MADA), separating the value of the resource from the value of disambiguation.
- Measuring how much systems disagree, plus an oracle upper bound, shows whether combining them is worth pursuing. Union and Intersection then trade recall against precision on demand.
- Breaking errors down by type (missed spans, truncated spans, wrong labels, per entity class) points directly to remedies such as gazetteers and span extension.

## For Mova

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

- **I1** (tokens): when English text loses its casing (headlines, all-caps, lowercased chat), the lexicon's proper-noun entries or a WordNet/Wikidata gloss can restore a capitalization-like feature for the tagger.
- **I2** (tokens): lexicon coverage in crate en can be exposed as an explicit, explainable feature for proper-noun tagging and for flagging candidate new entities.
- **I3** (learn): ablations should check whether a gain comes from a lexical resource itself or from choosing correctly among its options, which serves as a cleaner control in evaluation.
- **I4** (learn): when silver data is built from several witness parsers, pairwise agreement and an oracle score show whether witnesses are diverse enough, and intersection-style voting can produce high-precision silver labels.
- **I5** (world): entity extraction errors can be grouped the same way, and frequent misses on common places could be fixed by consulting Wikidata-backed entity lists in the world model.

*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

- [Columbia University](https://md.research.elamur.ai/free/en/universities/columbia-university/) — University, New York, United States · 241 works · 54,053 citations

Also on the byline: [Fair Isaac Corporation](https://ror.org/0354agh42).

📚 Shelves (4): [🖥️ Computer Science](https://md.research.elamur.ai/free/en/shelves/computer-science/) · 16,667 works · [🤖 Artificial Intelligence](https://md.research.elamur.ai/free/en/shelves/computer-science/artificial-intelligence/) · 16,194 works · [🎭 Arts and Humanities](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/) · 16,039 works · [🗣️ Language and Linguistics](https://md.research.elamur.ai/free/en/shelves/arts-and-humanities/language-and-linguistics/) · 15,891 works

🧭 Topics (3): [Semitic languages](https://md.research.elamur.ai/free/en/topics/semitic-languages/) · 2,908 works · [🏷️ Named entity recognition](https://md.research.elamur.ai/free/en/topics/named-entity-recognition/) · 200 works · [🧩 Morphology](https://md.research.elamur.ai/free/en/topics/morphology/) · 405 works

## Cite

Cite as: Benjamin Farber, Dayne Freitag, Nizar Habash and Owen Rambow. 2008. Improving NER in Arabic Using a Morphological Tagger. Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC'08).

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

```bibtex
@inproceedings{farber2008improving,
  title = {{Improving NER in Arabic Using a Morphological Tagger}},
  author = {Benjamin Farber, Dayne Freitag, Nizar Habash and Owen Rambow},
  booktitle = {Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC'08)},
  year = {2008},
  doi = {10.63317/47zx42oawm7v},
  url = {https://doi.org/10.63317/47zx42oawm7v}
}
```

- [← previous A Twin-Candidate Model for Learning-Based Anaphora Resolution](https://md.research.elamur.ai/free/en/papers/yang-2008-twin-candidate-model-learning-based/)
- [next → A Multi-Word Term Extraction Program for Arabic Language](https://md.research.elamur.ai/free/en/papers/boulaknadel-2008-multi-word-term-extraction-program/)
- [↑ back to papers of 2008](https://md.research.elamur.ai/free/en/papers/2008/)

## References

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

1. Shumeet Baluja, Vibhu Mittal, and Rahul Sukthankar. 1999. Applying machine learning for high performance named-entity extraction. In Pacific Association for Computational Linguistics. [↗](https://doi.org/10.1111/0824-7935.00129)
2. D.M. Bikel, S. Miller, R. Schwartz, and R. Weischedel. 1997. [Nymble: a high-performance learning name-finder](https://md.research.elamur.ai/free/en/papers/bikel-1997-nymble-high-performance-learning-name/). In Proc. 5th Conference on Applied Natural Language Processing (ANLP-97), April. [↗](https://doi.org/10.48550/arxiv.cmp-lg/9803003)
3. A. Borthwick. 1999. A Maximum Entropy Approach to Named Entity Recognition. Ph.D. thesis, New York University. [find](https://scholar.google.com/scholar?q=A%2E+Borthwick%2E+1999%2E+A+Maximum+Entropy+Approach+to+Named+Entity+Recognition%2E+Ph%2ED%2E+thesis%2C+New+York+University%2E)
4. Tim Buckwalter. 2004. Buckwalter Arabic Morphological Analyzer Version 2.0. Linguistic Data Consortium, University of Pennsylvania, 2002. LDC Cat alog No.: LDC2004L02, ISBN 1-58563-324-0. [↗](https://doi.org/10.35111/050q-5r95)
5. Michael Collins. 2002. [Discriminative training methods for hidden Markov models: theory and experiments with perceptron algorithms](https://md.research.elamur.ai/free/en/papers/collins-2002-discriminative-training-methods-hidden-markov/). In Proceedings of EMNLP 2002.
6. Dayne Freitag. 2004. Trained named entity recognition using distributional clusters. In Proceedings of EMNLP 2004. [find](https://scholar.google.com/scholar?q=Dayne+Freitag%2E+2004%2E+Trained+named+entity+recognition+using+distributional+clusters%2E+In+Proceedings+of+EMNLP+2004%2E)
7. Nizar Habash and Owen Rambow. 2005. [Arabic Tokenization, Part-of-Speech Tagging and Morphological Disambiguation in One Fell Swoop](https://md.research.elamur.ai/free/en/papers/habash-2005-arabic-tokenization-part-speech-tagging/). In Proceedings of the 43rd Annual Meeting of the Association for Computational Linguistics (ACL’05), pages 573–580, Ann Arbor, Michigan, June. Association for Computational Linguistics.
8. Nizar Habash, Abdelhadi Soudi, and Tim Buckwalter. 2007. On Arabic Transliteration. In A. van den Bosch and A. Soudi, editors, Arabic Computational Morphology: Knowledge-based and Empirical Methods. Springer. [↗](https://doi.org/10.1007/978-1-4020-6046-5_2)
9. Nizar Habash. 2007. Arabic Morphological Representations for Machine Translation. In A. van den Bosch and A. Soudi, editors, Arabic Computational Morphology: Knowledge-based and Empirical Methods. Springer. [↗](https://doi.org/10.1007/978-1-4020-6046-5_14)
10. H. Isozaki and H. Kazawa. 2002. Efficient support vector classifiers for named entity recognition. In Proceedings of COLING 2002. [find](https://scholar.google.com/scholar?q=H%2E+Isozaki+and+H%2E+Kazawa%2E+2002%2E+Efficient+support+vector+classifiers+for+named+entity+recognition%2E+In+Proceedings+of+COLING+2002%2E)
11. Mohamed Maamouri, Ann Bies, and Tim Buckwalter. 2004. The Penn Arabic Treebank: Building a large-scale annotated Arabic corpus. In NEMLAR Conference on Arabic Language Resources and Tools, Cairo, Egypt. [find](https://scholar.google.com/scholar?q=Mohamed+Maamouri%2C+Ann+Bies%2C+and+Tim+Buckwalter%2E+2004%2E+The+Penn+Arabic+Treebank%3A+Building+a+large%2Dscale+annotated+Arabic+corpus%2E+In+NEMLAR+Conference+on+Arabic+Language+Resources+and+Tools%2C+Cairo%2C+Egyp)
12. A. McCallum and W. Li. 2003. Early results for named entity recognition with conditional random fields. In Proceedings of the Seventh Conference on Natural Language Learning (CoNLL 2003). [find](https://scholar.google.com/scholar?q=A%2E+McCallum+and+W%2E+Li%2E+2003%2E+Early+results+for+named+entity+recognition+with+conditional+random+fields%2E+In+Proceedings+of+the+Seventh+Conference+on+Natural+Language+Learning+%28CoNLL+2003%29%2E)
13. S. Miller, J. Guinness, and A. Zamanian. 2004. [Name tagging with word clusters and discriminative training](https://md.research.elamur.ai/free/en/papers/miller-2004-name-tagging-word-clusters-discriminative/). In Proceedings of HLT/NAACL 04.

*Updated 1 November 2026*