Gulinigeer Abudouwaili

Ms. Gulinigeer Abudouwaili


Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
1 January 2023
Verified
Conference Paper
Open access
PDF file
GAGulinigeer Abudouwaili
Gulinigeer Abudouwaili
KAKahaerjiang Abiderexiti
Kahaerjiang Abiderexiti
NYNian Yi
Nian Yi
AWAishan Wumaier
Aishan Wumaier
Gulinigeer Abudouwaili, Kahaerjiang Abiderexiti, Nian Yi, Aishan Wumaier. Proceedings of the 20th SIGMORPHON workshop on Computational Research in Phonetics, Phonology, and Morphology. 2023.

10 February 2025 in Complex & Intelligent Systems
Verified
Journal Article
Open access
PDF file
GAGulinigeer Abudouwaili
Gulinigeer Abudouwaili
SRSirajahmat Ruzmamat
Sirajahmat Ruzmamat
KAKahaerjiang Abiderexiti
Kahaerjiang Abiderexiti
TYTuergen Yibulayin
Tuergen Yibulayin
NYNian Yi
Nian Yi
AWAishan Wumaier
Aishan Wumaier
Data sparsity and out-of-vocabulary are the main challenges in low-resource machine translation, and the impact of such problems in translation can be reduced through word segmentation. Word segmentation can be roughly divided into two categories: unsupervised word segmentation and morphological knowledge-based word segmentation. However, the performance of the two methods when applied to downstream tasks varies, and the effectiveness of this method for out-of-vocabulary problem has not yet been verified. Therefore, we first explore the impact of mainstream subword segmentation methods on machine translation and analyze the advantages and disadvantages of the two methods. Secondly, we utilized the advantages of the two methods to improve the existing segmentation methods. Two semi-supervised subword segmentation methods combining morphological knowledge were proposed. The impact of these methods on the distributions of vocabulary and out-of-vocabulary words was analyzed. Once again, a dual encoder was introduced in the encoding to improve the model’s ability to extract information. We improved the existing information fusion methods during feature fusion to avoid information loss. Finally, validation experiments were conducted on low-resource agglutinative languages to Chinese machine translation using Kazakh, Uyghur, and Uzbek as case studies. In this work, we proposed a new subword segmentation method and a novelty machine translation model to improve translation quality to varying degrees, especially on lower data resources.

21 June 2024 in Applied Sciences
Verified
Journal Article
Open access
PDF file
GAGulinigeer Abudouwaili
Gulinigeer Abudouwaili
SRSirajahmat Ruzmamat
Sirajahmat Ruzmamat
KAKahaerjiang Abiderexiti
Kahaerjiang Abiderexiti
BWBinghong Wu
Binghong Wu
AWAishan Wumaier
Aishan Wumaier
Morphological segmentation and stemming are foundational tasks in natural language processing. They have become effective ways to alleviate data sparsity in agglutinative languages because of the nature of agglutinative language word formation. Uyghur and Kazakh, as typical agglutinative languages, have made significant progress in morphological segmentation and stemming in recent years. However, the evaluation metrics used in previous work are character-level based, which may not comprehensively reflect the performance of models in morphological segmentation or stemming. Moreover, existing methods avoid manual feature extraction, but the model’s ability to learn features is inadequate in complex scenarios, and the correlation between different features has not been considered. Consequently, these models lack representation in complex contexts, affecting their effective generalization in practical scenarios. To address these issues, this paper redefines the morphological-level evaluation metrics: F1-score and accuracy (ACC) for morphological segmentation and stemming tasks. In addition, two models are proposed for morpheme segmentation and stem extraction tasks: supervised model and unsupervised model. The supervised model learns character and contextual features simultaneously, then feature embeddings are input into a Transformer encoder to study the correlation between character and context embeddings. The last layer of the model uses a CRF or softmax layer to determine morphological boundaries. In unsupervised learning, an encoder–decoder structure introduces n-gram correlation assumptions and masked attention mechanisms, enhancing the correlation between characters within n-grams and reducing the impact of characters outside n-grams on boundaries. Finally, comprehensive comparative analyses of the performance of different models are conducted from various points of view. Experimental results demonstrate that: (1) The proposed evaluation method effectively reflects the differences in morphological segmentation and stemming for Uyghur and Kazakh; (2) Learning different features and their correlation can enhance the model’s generalization ability in complex contexts. The proposed models achieve state-of-the-art performance on Uyghur and Kazakh datasets.