Dataset Card for "Masked Arab States Dataset (MASD)"

This dataset is created using 20 Arab States¹ with their corresponding capital cities, nationalities, currencies, and on which continents they are located, consisting of four categories: country-capital prompts, country-currency prompts, country-nationality prompts, and country-continent prompts. Each prompts category has 40 masked prompts, and the total number of masked prompts in the MASD dataset is 160. This dataset is used to evaluate these Arabic Masked Language Models (MLMs):

1. SaiedAlshahrani/arwiki_20230101_roberta_mlm_bots.
2. SaiedAlshahrani/arwiki_20230101_roberta_mlm_nobots.
3. SaiedAlshahrani/arzwiki_20230101_roberta_mlm.
4. SaiedAlshahrani/arywiki_20230101_roberta_mlm_bots.
5. SaiedAlshahrani/arywiki_20230101_roberta_mlm_nobots.

For more details about the dataset, please read and cite our paper:

@inproceedings{alshahrani-etal-2023-performance,
    title = "{Performance Implications of Using Unrepresentative Corpora in {A}rabic Natural Language Processing}",
    author = "Alshahrani, Saied  and Alshahrani, Norah  and Dey, Soumyabrata  and Matthews, Jeanna",
    booktitle = "Proceedings of the The First Arabic Natural Language Processing Conference (ArabicNLP 2023)",
    month = December,
    year = "2023",
    address = "Singapore (Hybrid)",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2023.arabicnlp-1.19",
    doi = "10.18653/v1/2023.arabicnlp-1.19",
    pages = "218--231",
    abstract = "Wikipedia articles are a widely used source of training data for Natural Language Processing (NLP) research, particularly as corpora for low-resource languages like Arabic. However, it is essential to understand the extent to which these corpora reflect the representative contributions of native speakers, especially when many entries in a given language are directly translated from other languages or automatically generated through automated mechanisms. In this paper, we study the performance implications of using inorganic corpora that are not representative of native speakers and are generated through automated techniques such as bot generation or automated template-based translation. The case of the Arabic Wikipedia editions gives a unique case study of this since the Moroccan Arabic Wikipedia edition (ARY) is small but representative, the Egyptian Arabic Wikipedia edition (ARZ) is large but unrepresentative, and the Modern Standard Arabic Wikipedia edition (AR) is both large and more representative. We intrinsically evaluate the performance of two main NLP upstream tasks, namely word representation and language modeling, using word analogy evaluations and fill-mask evaluations using our two newly created datasets: Arab States Analogy Dataset (ASAD) and Masked Arab States Dataset (MASD). We demonstrate that for good NLP performance, we need both large and organic corpora; neither alone is sufficient. We show that producing large corpora through automated means can be a counter-productive, producing models that both perform worse and lack cultural richness and meaningful representation of the Arabic language and its native speakers.",
}

_{1. We only drop two Arab states: the United Arab Emirates (الإمارات العربية المتحدة) and Comoros (جزر القمر), because they or their capital cities are written as open compound words (two words), which cannot be directly handled by the word embedding models, like Abu Dhabi (أبو ظبي).}