Elamur Research⌕
☰
home / papers / 2016 / Al Qamus al Muhit, a Medieval Arabic Lexicon in LMF

Al Qamus al Muhit, a Medieval Arabic Lexicon in LMF

Paper · 2016 · Ouafae Nahli, Francesca Frontini, Monica Monachini et al.

Summary

The paper tackles how to make a classical Arabic dictionary usable by computers. The authors take ’al-qāmūs al-muḥīṭ (AQAM), the comprehensive and highly authoritative lexicon compiled by al-Fīrūzābādī (1329–1414), and convert it into LMF (Lexical Markup Framework), the ISO standard for NLP lexicons. They start from a digitized plain-text version that they checked against a printed 1998 Beirut edition. They manually work out its markup conventions, both explicit and implicit, and then segment the text in two passes: first into whole entries, then into lemmata and definitions. The entries are encoded in XML following the LMF NLP module, using earlier Arabic LMF examples (Khemakhem 2013) as a model. A single dictionary entry expands into several LMF entries covering the verb, its masdars and a derived adjective. Each sense can carry relations such as antonymy and links to Princeton WordNet and SUMO. The theoretical result is an argument that consonantal roots should not be encoded as lexical entries. Instead, they should be treated as classifying metadata that tie derived words together. The finished resource would serve Arabic NLP, particularly for older texts.

Key points

  • AQAM is ordered by consonantal root in three levels: the root's last consonant defines 28 chapters, the first consonant defines sub-chapters, and the second consonant orders entries within those.
  • The source text uses explicit markers such as *1* and *2* for the third and first radical. It also uses implicit ones, such as @, which signals that the second radical has changed without saying which consonant it is now.
  • Segmentation is coarse-to-fine. Blocks are first labelled "Unsegmented Text" and later divided into lemma sets and definitions. Commas and colons separate the morphological material from the lexical material.
  • In the worked example kabura ("grow; be great"), the verb has scheme fa'ula and inflectional paradigm u/u. It yields three masdar entries and an adjective kabīr with two plural WordForms, and every entry is linked to the root kbr through RelatedForm.
  • Senses are connected outward with MonolingualExternalRef entries of type near_syn, pointing to PWN3.1 (200125649v) and the SUMO concept FullyFormed. The antonym ṣaġura is linked through a SenseRelation.

Interesting ideas

  • I1. The authors exploit regularities in the source: a verb of scheme fa'ula reliably comes with a fa'īl adjective introduced by "fa=huwa", so whole sets of related forms can be generated by rule instead of annotated by hand.
  • I2. A root is treated as a key that groups lexemes, not as a lexeme in its own right, because it has no precise meaning of its own and counting it as a word would inflate the entry count.
  • I3. Text that has not yet been parsed is kept under an explicit "Unsegmented Text" tag and refined in a later pass, rather than being forced into structure right away.
  • I4. Links to external resources are typed as near_syn rather than exact equivalence, which states openly that a dictionary sense and a WordNet synset or SUMO concept may only partly overlap.
  • I5. One condensed dictionary entry is expanded into several standard entries, each with its own scheme, forms and senses, which turns the dictionary's implicit structure into explicit records.

For Mova

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

  • I1 (tokens): Regular English derivation pairs, such as verb to -tion noun or adjective to -ness noun, could be generated by rule and stored as explicit links. Frames could then map a nominalization back to its verb's roles.
  • I2 (world): Word families or derivational stems could be stored as attributes on lemma nodes in the ontology, so they never appear as fake entities with senses of their own.
  • I3 (learn): When silver data is built, spans the parsers cannot analyse could be marked as unresolved residue. This keeps them out of the training data while still recording them for later repair rules.
  • I4 (world): Mappings between VerbNet, WordNet, FrameNet and Wikidata could carry an exactness label, so reasoning can tell what is firm and explanations show when a link is only approximate.
  • I5 (world): Compact compiled knowledge, such as verb class entries, could be expanded into separate explicit nodes for each related form. Every derived fact would then be traceable to the source line it came from.

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

📚 Shelves 5 Computer Science · 9,086 works Artificial Intelligence · 8,889 works Arts and Humanities · 9,296 works Language and Linguistics · 9,206 works Literature and Literary Theory · 340 works

🧭 Topics 1 Semitic languages · 1,484 works

Cite

Cite as Ouafae Nahli, Francesca Frontini, Monica Monachini, Fahad Khan, Arsalan Zarghili and Mustapha Khalfi. 2016. Al Qamus al Muhit, a Medieval Arabic Lexicon in LMF. Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16).

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

@inproceedings{nahli2016al,
  title = {{Al Qamus al Muhit, a Medieval Arabic Lexicon in LMF}},
  author = {Ouafae Nahli, Francesca Frontini, Monica Monachini, Fahad Khan, Arsalan Zarghili and Mustapha Khalfi},
  booktitle = {Proceedings of the Tenth International Conference on Language Resources and Evaluation (LREC'16)},
  year = {2016},
  doi = {10.63317/4tw7vhnx5xco},
  url = {https://doi.org/10.63317/4tw7vhnx5xco}
}

References

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

  1. Attia, M., Toral, A., Tounsi, L., Monachini, M., and van Genabith, J. (2010). An automatically built named entity lexicon for arabic. In Nicoletta Calzolari (Conference Chair), et al., editors, Proceedings of the Seventh International Conference on Language Resources and Evaluation (LREC’10), Valletta, Malta, may. European Language Resources Association (ELRA). find
  2. 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
  3. Cantineau, J. (1950). Racines et schémes. In Mélanges offerts á William Marais par l’institut d’études islamiques de l’Université de Paris. GP Maisonneuve, Paris. find
  4. Carter, M. (1990). Arabic lexicography. In J. D. Latham M. J. L. Youg et al., editors, Learning and Science in the Abbasid Period (“The Cambridge History of Arabic Literature”), pages 106–117. Cambridge University Press. find
  5. Del Grosso, A. M. and Nahli, O. (October 2014). Towards a flexible open-source software library for multi-layered scholarly textual studies: An Arabic case study dealing with semi-automatic language processing. In Proceedings of 3rd IEEE International Colloquium, Information Science and Technology (CIST),Tetouan, Marocco, pages 285–290, Washington, DC, USA. IEEE. find
  6. Dichy, J. (2002). Sens des schèmes et sens des racines en arabe : le principe de figement lexical (pfl) et ses effets sur le vocabulaire d’une langue sémitique. In L. Panier et al., editors, La polysémie. Presses Universitaires de Lyon. find
  7. Elkateb, S., Black, W., Rodrı́guez, H., Alkhalifa, M., Vossen, P., Pease, A., and Fellbaum, C. (2006). Building a wordnet for arabic. In Proceedings of The fifth international conference on Language Resources and Evaluation (LREC 2006). Citeseer. find
  8. Fellbaum, C. (1998). WordNet. Wiley Online Library. tain senses or synsets from different languages (Francopoulo and George, 2008). Figure 4: Partial schema of the LMF version of the lexicon. find
  9. Francopoulo, G. and George, M. (2008). Language Resource Management. Lexical markup framework (LMF). Technical report, ISO/TC 37/SC 4 N453 (N330 Rev. 16. our card
  10. G. Francopoulo, editor. (2013). LMF Lexical Markup Framework. John Wiley & Sons. find
  11. Khemakhem, A., Gargouri, B., Abdelwahed, A., and Francopoulo, G. (2007). Modélisation des paradigmes de flexion des verbes arabes selon la norme LMF-ISO 24613 . In Proceedings of TALN. find
  12. Khemakhem, A., Elleuch, I., Gargouri, B., and Hamadou, A. B. (2009). Towards an automatic conversion approach of editorial Arabic dictionaries into LMF-ISO 24613 standardized model. In Proceedings of the Second International Conference on Arabic Language Resources and Tools, Cairo, Egypt. find
  13. Khemakhem, A., Gargouri, B., Haddar, K., and Ben-Hamadou, A. (2013). Lmf for arabic. In G. Francopoulo, editor, LMF Lexical Markup Framework, pages 83–98. John Wiley & Sons. ↗
  14. Kouloughli, D. E. (1991). Grammaire de l’arabe d’aujourdhui. Édition Pocket. find
  15. Lancioni, G. (1997). Sull’ordinamento dei dizionari arabi classici. Rivista degli studi orientali, 71:113–127. find
  16. Loukil, N. and Haddar, K. ahd BenHamadou, A. (2007). Normalisation de la représentation des lexiques syntaxiques arabes pour les formalismes dunification. In Proceedings of the 26th conference on Lexis and Grammar, Bonifacio. find
  17. A. Moutaouakil, editor. (1989). Pragmatic Functions in a Functional Grammar of Arabic. Foris Publications - Holland. find
  18. Pease, A., Niles, I., and Li, J. (2002). The suggested upper merged ontology: A large ontology for the semantic web and its applications. In Working notes of the AAAI-2002 workshop on Ontologies and the Semantic Web, volume 28. find
  19. Rodrı́guez, H., Farwell, D., Farreres, J., Bertran, M., Alkhalifa, M., Martı́, M. A., Black, W., Elkateb, S., Kirk, J., Pease, A., Piek, V., and Fellbaum, C. (2008a). Arabic wordnet: Current state and future extensions. In Proceedings of the Fourth International GlobalWordNet Conference - GWC 2008, Szeged, Hungary, January 22-25. find
  20. Rodrı́guez, H., Farwell, D., Ferreres, J., Bertran, M., Alkhalifa, M., and Martı́, M. A. (2008b). Arabic wordnet: Semi-automatic extensions using bayesian inference. In Proceedings of the the 6th Conference on Language Resources and Evaluation (LREC 2008). Marrakech (Morocco). find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/nahli-2016-al-qamus-al-muhit-medieval/