Elamur Research⌕
☰
home / papers / 2020 / Developing an Arabic Infectious Disease Ontology to Include Non-Standard Terminology

Developing an Arabic Infectious Disease Ontology to Include Non-Standard Terminology

Summary

The paper addresses the near absence of medical ontologies for Arabic, and specifically the lack of any infectious disease ontology that accounts for how ordinary people talk about illness, for example on social media. The authors build the resource with a hybrid pipeline that combines automatic ontology learning with manual curation. They first crawled trusted government, organisational and hospital websites, guided by a WHO-based list of diseases common in Arab countries. This produced a corpus of 115 documents on about 25 diseases. The text was then tokenized, normalized, stripped of stopwords, POS-tagged and sentence-split. Candidate terms were mined with three statistical extractors: TF-IDF, C-value and YAKE. To capture lay vocabulary, the authors consulted roughly 100 people and linked the colloquial expressions they reported to the scientific concepts. Relations between concepts were created by hand for now. Evaluation has two parts: per-disease precision, recall and F-measure for term extraction, and a qualitative review by a paediatric infectious disease consultant. The resource matters because it supports disease surveillance over informal Arabic text, where dialect and slang defeat tools built only for Modern Standard Arabic.

Key points

  • Corpus: 115 documents covering around 25 infectious diseases, crawled only from trusted sources (governments, organisations, hospitals), with the disease list derived from WHO information.
  • Term extraction: three methods were compared.
    • TF-IDF over uni-, bi- and trigrams via scikit-learn 0.20.2, discarding terms with document frequency below two.
    • C-value, aimed at nested multi-word terms, applied on top of the Stanford Arabic POS tagger with a linguistic filter.
    • YAKE, an unsupervised keyword extractor.
  • Informal terms: a study with about 100 participants collected lay and slang terms, which were then mapped onto formal disease vocabulary.
  • Shortcomings of BioCaster: the existing multilingual ontology is translation-based. It can mistranslate Arabic disease names (meningitis rendered as schizophrenia) and mishandles English abbreviations such as MRSA.
  • Evaluation: extraction scores are reported per disease (Table 4) and mostly fall in a modest range of about 20–50%. A domain expert reviewed the ontology qualitatively. Relations are still manual, and automatic relation extraction is left for future work.

Interesting ideas

  • I1. Each scientific concept is stored together with its colloquial and dialectal surface forms, so text in either register resolves to the same node.
  • I2. Several term extractors with different biases are run side by side: frequency-based (TF-IDF), nested multi-word aware (C-value) and feature-based keyword scoring (YAKE). Each is then scored against a reference list.
  • I3. Lay vocabulary is gathered by directly asking about a hundred speakers rather than relying only on corpora, which finds everyday expressions that formal documents never contain.
  • I4. The paper shows that building a multilingual ontology by translating labels breaks on compound disease names and abbreviations. Labels should be native per language, or abbreviations expanded before mapping.
  • I5. Evaluation combines an automatic metric with a qualitative review by a domain expert, which catches errors that numbers alone miss.

For Mova

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

  • I1 (world): Our ontology layer could attach informal synonyms ("tummy bug", "the flu") to canonical WordNet/Wikidata concepts, letting the world model ground casual text without guessing.
  • I2 (learn): Disagreement between such extractors could act as witnesses for silver multi-word lexicon entries, with only terms agreed by several methods promoted into the lexicon.
  • I3 (world): A small elicitation round, from people or LLM teacher tiers prompted as lay speakers, could fill gaps in informal verb and noun coverage for the knowledge graph.
  • I4 (back): When translating through the meta-world, we should realise concept nodes from per-language lexicalisations instead of translating source strings, and expand acronyms at the graph level.
  • I5 (learn): Ontology or frame additions in specialised domains could pair automatic precision/recall checks with a small expert-reviewed sample as a sanity control.

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: Linguistics & languages

Topics Knowledge graphs & ontologies Semitic languages

 Institutions

🎓Lancaster UniversityLancaster · United Kingdom13 works · 1,069 citations

Cite

Cite as Lama Alsudias and Paul Rayson. 2020. Developing an Arabic Infectious Disease Ontology to Include Non-Standard Terminology. Proceedings of the Twelfth Language Resources and Evaluation Conference.

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

@inproceedings{alsudias2020developing,
  title = {{Developing an Arabic Infectious Disease Ontology to Include Non-Standard Terminology}},
  author = {Lama Alsudias and Paul Rayson},
  booktitle = {Proceedings of the Twelfth Language Resources and Evaluation Conference},
  year = {2020},
  url = {https://aclanthology.org/2020.lrec-1.596/}
}

References

21 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Al-Safadi, L., Al-Badrani, M., and Al-Junidey, M. (2011). Developing ontology for arabic blogs retrieval. International Journal of Computer Applications, 19(4):40–45. find
  2. Al-Thubaity, A. M., Khan, M., Alotaibi, S., and Alonazi, B. (2014). Automatic arabic term extraction from special domain corpora. In 2014 International Conference on Asian Language Processing (IALP), pages 1–5. IEEE. find
  3. Al-Zoghby, A. M., Elshiwi, A., and Atwan, A., (2018). Semantic Relations Extraction and Ontology Learning from Arabic Texts—A Survey, pages 199–225. Springer International Publishing, Cham. find
  4. AlAgha, I. M. and Abu-Taha, A. (2015). Ar2sparql: an arabic natural language interface for the semantic web. AR2SPARQL: An Arabic Natural Language Interface for the Semantic Web, 124(18). find
  5. Albukhitan, S. and Helmy, T. (2013). Automatic ontology-based annotation of food, nutrition and health arabic web content. Procedia Computer Science, 19:461–469. find
  6. Alsudias, L. and Rayson, P. (2019). Classifying information sources in Arabic twitter to support online monitoring of infectious diseases. In Proceedings of the 3rd Workshop on Arabic Corpus Linguistics, pages 22–30, Cardiff, United Kingdom, 22 July. Association for Computational Linguistics. find
  7. Asim, M. N., Wasim, M., Khan, M. U. G., Mahmood, W., and Abbasi, H. M. (2018). A survey of ontology learning techniques and applications. find
  8. Black, W., Elkateb, S., Rodriguez, H., Alkhalifa, M., Vossen, P., Pease, A., and Fellbaum, C. (2006). Introducing the arabic wordnet project. In Proceedings of the third international WordNet conference, pages 295–300. Citeseer. find
  9. Campos, R., Mangaravite, V., Pasquali, A., Jorge, A., Nunes, C., and Jatowt, A. (2019). Yake! keyword extraction from single documents using multiple local features. Information Sciences. find
  10. Cimiano, P. (2006). Ontology learning from text. Ontology Learning and Population from Text: Algorithms, Evaluation and Applications, pages 19–34. find
  11. Collier, N., Goodwin, R. M., McCrae, J., Doan, S., Kawazoe, A., Conway, M., Kawtrakul, A., Takeuchi, K., and Dien, D. (2010). An ontology-driven system for detecting global health events. In Proceedings of the 23rd International Conference on Computational Linguistics, pages 215–222. Association for Computational Linguistics. find
  12. Cowell, L. G. and Smith, B. (2010). Infectious disease ontology. In Infectious disease informatics, pages 373– 395. Springer. find
  13. Frantzi, K., Ananiadou, S., and Mima, H. (2000). Automatic recognition of multi-word terms: the c-value/nc-value method. International journal on digital libraries, 3(2):115–130. find
  14. Joshi, A., Karimi, S., Sparks, R., Paris, C., and Macintyre, C. R. (2019). Survey of text-based epidemic intelligence: A computational linguistics perspective. ACM Comput. Surv., 52(6):119:1–119:19, October. find
  15. Kapoor, B. and Sharma, S. (2010). A comparative study ontology building tools for semantic web applications. International Journal of Web & Semantic Technology (IJWesT), 1(3):1–13. find
  16. Manning, C., Surdeanu, M., Bauer, J., Finkel, J., Bethard, S., and McClosky, D. (2014). The Stanford CoreNLP natural language processing toolkit. In Association for Computational Linguistics (ACL) System Demonstrations, pages 55–60. our card
  17. Musen, M. A. et al. (2015). The protégé project: a look back and a look forward. AI matters, 1(4):4. find
  18. Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion, B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R., Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D., Brucher, M., Perrot, M., and Duchesnay, E. (2011). Scikit-learn: Machine learning in Python. Journal of Machine Learning Research, 12:2825–2830. find
  19. Raisan, M. and Abdullah, A. M. (2017). Building a Core Arabic Ontology for Iraqi News. Noor Publishing. find
  20. Schriml, L. M., Arze, C., Nadendla, S., Chang, Y.-W. W., Mazaitis, M., Felix, V., Feng, G., and Kibbe, W. A. (2011). Disease ontology: a backbone for disease semantic integration. Nucleic acids research, 40(D1):D940–D946. find
  21. Soliman, T. H., Elmasry, M., Hedar, A., and Doss, M. (2014). Sentiment analysis of arabic slang comments on facebook. International Journal of Computers & Technology, 12(5):3470–3478. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/alsudias-2020-developing-arabic-infectious-disease-ontology/