We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:


References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Computation and Language

Title: Adapting MARBERT for Improved Arabic Dialect Identification: Submission to the NADI 2021 Shared Task

Abstract: In this paper, we tackle the Nuanced Arabic Dialect Identification (NADI) shared task (Abdul-Mageed et al., 2021) and demonstrate state-of-the-art results on all of its four subtasks. Tasks are to identify the geographic origin of short Dialectal (DA) and Modern Standard Arabic (MSA) utterances at the levels of both country and province. Our final model is an ensemble of variants built on top of MARBERT that achieves an F1-score of 34.03% for DA at the country-level development set -- an improvement of 7.63% from previous work.
Comments: This work was accepted at the Sixth Arabic Natural Language Processing Workshop (EACL/WANLP 2021)
Subjects: Computation and Language (cs.CL)
Cite as: arXiv:2103.01065 [cs.CL]
  (or arXiv:2103.01065v1 [cs.CL] for this version)

Submission history

From: Badr AlKhamissi [view email]
[v1] Mon, 1 Mar 2021 15:19:56 GMT (801kb,D)

Link back to: arXiv, form interface, contact.