Elamur Research⌕
☰
home / papers / 2014 / AIDArabic A Named-Entity Disambiguation Framework for Arabic Text

AIDArabic A Named-Entity Disambiguation Framework for Arabic Text

Paper · 2014 · Mohamed Amir Yosef, Marc Spaniol and Gerhard Weikum

Summary

The paper addresses named-entity disambiguation (NED) for Arabic. NED means linking an ambiguous name in text to the right entry in a knowledge base. Arabic resources are thin: in July 2014 the Arabic Wikipedia had under 0.3 million articles, against more than 4.5 million in English. That leaves gaps in entity coverage, anchor texts, categories and link structure.

The authors start from their open-source English system AIDA and rework its data pipeline. They use the multilingual knowledge base YAGO3, add Arabic names from Freebase, and follow interwiki links between the Arabic and English Wikipedias. This lets an Arabic mention resolve to an entity that exists only on the English side. It also enriches each entity's description and the relatedness model with English-side evidence. They deliberately avoid machine translation to keep the data precise.

Their proof-of-concept indicates that Arabic NED is workable despite sparse sources. Cross-lingual fusion supplies more context than an Arabic-only approach. The remaining errors come mainly from Arabic clitics, which stop names and keywords from matching the dictionary and entity descriptions.

Key points

  • The paper splits an NED framework into four ingredients: an entity repository, a many-to-many name–entity dictionary, weighted keyphrase descriptions per entity, and an entity–entity relatedness model.
  • YAGO3, the multilingual successor of YAGO2s, replaces the earlier repository. Freebase is used only for its multilingual names, with Freebase entities mapped onto YAGO3 entities.
  • Extractors produce an Entity Dictionary from Arabic–English interwiki links and a Categories Dictionary mapping categories between the two Wikipedias.
  • Relatedness is computed over the merged link graphs of the English and Arabic Wikipedias. Entity context draws on anchor texts, categories and links, with no automated translation.
  • Clitics hurt recall: attached prepositions and pronouns block candidate lookup and keyword matching. Planned future work includes a proper evaluation dataset, translation or transliteration of keyphrases, and approximate matching.

Interesting ideas

  • I1. When one resource is sparse, align it to a richer one through explicit identity links and borrow that resource's descriptions and link structure, rather than translating text.
  • I2. Model disambiguation as four separable resources (catalog, name dictionary, scored descriptions, relatedness), so each part can be swapped or improved independently.
  • I3. Refusing machine translation for entity context trades recall for precision and keeps noisy derived evidence out of the knowledge base.
  • I4. Coherence is enforced jointly: mentions and candidate entities form one graph, which is pruned until each mention keeps one entity that fits with the others.
  • I5. Lookup failures caused by attached affixes show that dictionary and description matching must run on normalized forms.

For Mova

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

  • I1 (world): Rare entities in Mova's knowledge graph could inherit context from Wikidata and WordNet entries linked by IDs, keeping a recorded provenance for each borrowed fact.
  • I2 (world): An entity-linking layer built this way stays explainable, because every link decision can be traced to a dictionary hit, a description match, or a coherence score.
  • I3 (learn): Silver knowledge could be admitted only when it comes from aligned sources, never from generated text, with a negative control checking that skipped noisy sources do not secretly improve accuracy.
  • I4 (frames): Coreference and entity resolution in a story could pick referents that are mutually consistent across the whole text, not one mention at a time.
  • I5 (tokens): Gazetteer and keyphrase lookup should strip possessive 's, hyphenation and inflection via lemmas before matching, so names like "Cairo's" still find candidates.

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

🔬Max Planck Institute for InformaticsSaarbrücken · Germany16 works · 4,453 citations🔬Max Planck SocietyMunich · Germany16 works · 42,253 citations

 Fields, topics and concepts

📚 Shelves 6The 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 worksDecision Sciences · 209 worksManagement Science and Operations Research · 108 works

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

Cite

Cite as Mohamed Amir Yosef, Marc Spaniol and Gerhard Weikum. 2014. AIDArabic A Named-Entity Disambiguation Framework for Arabic Text. 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/w14-3626

@inproceedings{yosef2014aidarabic,
  title = {{AIDArabic A Named-Entity Disambiguation Framework for Arabic Text}},
  author = {Mohamed Amir Yosef, Marc Spaniol and Gerhard Weikum},
  booktitle = {Proceedings of the EMNLP 2014 Workshop on Arabic Natural Language Processing (ANLP)},
  year = {2014},
  doi = {10.3115/v1/w14-3626},
  url = {https://doi.org/10.3115/v1/w14-3626}
}

References

13 references, 3 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Razvan Bunescu and Marius Pasca. 2006. Using encyclopedic knowledge for named entity disambiguation. In Proceedings of the 11th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2006), pages 9–16, Trento, Italy.
  2. S. Cucerzan. 2007. Large-scale named entity disambiguation based on Wikipedia data. In Proceedings of EMNLP-CoNLL 2007, pages 708–716, Prague, Czech Republic. find
  3. 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 (ACL 2013) , pages 1558–1567, Sofia, Bulgaria.
  4. Paolo Ferragina and Ugo Scaiella. 2010. Tagme: On-the-fly annotation of short text fragments (by Wikipedia entities). In Proceedings of the 19th ACM International Conference on Information and Knowledge Management (CIKM 2010), pages 1625–1628, New York, NY, USA.
  5. D. A. Ferrucci. 2012. Introduction to ”This is Watson”. IBM Journal of Research and Development (Volume 56, Issue 3), pages 235–249. ↗
  6. Johannes Hoffart, Mohamed Amir Yosef, Ilaria Bordino, Hagen Fürstenau, Manfred Pinkal, Marc Spaniol, Bilyana Taneva, Stefan Thater, and Gerhard Weikum. 2011. Robust disambiguation of named entities in text. In Proceedings of the Conference on Empirical Methods in Natural Language Processing (EMNLP 2011), pages 782– 792, Edinburgh, Scotland. find
  7. Johannes Hoffart, Fabian M. Suchanek, Klaus Berberich, and Gerhard Weikum. 2013. YAGO2: A spatially and temporally enhanced knowledge base from Wikipedia. Artificial Intelligence (Volume 194), pages 28–61. ↗
  8. Sayali Kulkarni, Amit Singh, Ganesh Ramakrishnan, and Soumen Chakrabarti. 2009. Collective annotation of Wikipedia entities in web text. In Proceedings of the 15th ACM International Conference on Knowledge Discovery and Data Mining (SIGKDD 2009), pages 457–466, New York, NY, USA. find
  9. Farzane Mahdisoltani, Joanna Biega, and Fabian M. Suchanek. 2014. A knowledge base from multilingual Wikipedias [Yosef et al.2011] Mohamed Amir Yosef, Johannes Hoffart, Ilaria Bordino, Marc Spaniol, and Gerhard – yago3. Technical report, Telecom ParisTech. Weikum. 2011. AIDA: An online tool for accu-http://suchanek.name/work/publications/yago3tr.pdf. rate disambiguation of named entities in text and tables. In Proceedings of the 37th International Con- ↗
  10. Pablo N. Mendes, Max Jakob, ference on Very Large Data Bases (VLDB 2011), Andrés Garcı́a-Silva, and Christian Bizer. 2011. pages 1450–1453, Seattle, WA, USA. DBbpedia Spotlight: Shedding light on the web of documents. In Proceedings of the 7th International Conference on Semantic Systems ( I-Semantics 2011), pages 1–8, New York, NY, USA. find
  11. David N. Milne and Ian H. Witten. 2008. Learning to link with Wikipedia. In Proceedings of the 17th ACM International Conference on Information and Knowledge Management (CIKM 2008), pages 509–518, New York, NY, USA. find
  12. Hien T. Nguyen and Tru H. Cao. 2008. Named entity disambiguation on an ontology enriched by Wikipedia. In Proceedings of IEEE International Conference on Research, Innovation and Vision for the Future (RIVF 2008), pages 247–254, Ho Chi Minh City, Vietnam. find
  13. Lev Ratinov, Dan Roth, Doug Downey, and Mike Anderson. 2011. Local and global algorithms for disambiguation to wikipedia. In Proceedings of the 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies (HLT 2011), pages 1375– 1384, Stroudsburg, PA, USA. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/yosef-2014-aidarabic-named-entity-disambiguation-framework/