Research on the Uyghur morphological segmentation model with an attention mechanism
Morphological segmentation is a basic task in agglutinative language information processing, dividing words into the smallest semantic unit morphemes. There are two types of morphological segmentation: canonical segmentation and surface segmentation. As a typical agglutinative language, Uyghur usual...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Taylor & Francis Group
2022-12-01
|
| Series: | Connection Science |
| Subjects: | |
| Online Access: | http://dx.doi.org/10.1080/09540091.2022.2134843 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849395460442161152 |
|---|---|
| author | Gulinigeer Abudouwaili Kahaerjing Abiderexiti Yunfei Shen Aishan Wumaier |
| author_facet | Gulinigeer Abudouwaili Kahaerjing Abiderexiti Yunfei Shen Aishan Wumaier |
| author_sort | Gulinigeer Abudouwaili |
| collection | DOAJ |
| description | Morphological segmentation is a basic task in agglutinative language information processing, dividing words into the smallest semantic unit morphemes. There are two types of morphological segmentation: canonical segmentation and surface segmentation. As a typical agglutinative language, Uyghur usually uses statistical-based methods in canonical segmentation, which relies on the artificial extraction of features. In surface segmentation, the artificial feature extraction process is avoided by using the neural network. However, to date, no model can provide both segmentation results in Uyghur without adding features. In addition, morphological segmentation is usually regarded as a sequence annotation task, so label imbalance easily occurs in datasets. Given the above situation, this paper proposes an improved labelling scheme that joins morphological boundary labels and voice harmony labels for the two kinds of segmentation simultaneously. Then, a convolution network and attention mechanism are added to capture local and global features, respectively. Finally, morphological segmentation is regarded as a sequence labeling task of character sequences. Due to the problem of label proportion imbalance and noise in the dataset, a focal loss function with label smoothing is used. The experimental results show that the F1 values of canonical segmentation and surface segmentation achieve the best results. |
| format | Article |
| id | doaj-art-a78ddf8f05a44086b4b0f4110e976bcb |
| institution | Kabale University |
| issn | 0954-0091 1360-0494 |
| language | English |
| publishDate | 2022-12-01 |
| publisher | Taylor & Francis Group |
| record_format | Article |
| series | Connection Science |
| spelling | doaj-art-a78ddf8f05a44086b4b0f4110e976bcb2025-08-20T03:39:36ZengTaylor & Francis GroupConnection Science0954-00911360-04942022-12-013412577259610.1080/09540091.2022.21348432134843Research on the Uyghur morphological segmentation model with an attention mechanismGulinigeer Abudouwaili0Kahaerjing Abiderexiti1Yunfei Shen2Aishan Wumaier3Xinjiang UniversityXinjiang UniversityXinjiang UniversityXinjiang UniversityMorphological segmentation is a basic task in agglutinative language information processing, dividing words into the smallest semantic unit morphemes. There are two types of morphological segmentation: canonical segmentation and surface segmentation. As a typical agglutinative language, Uyghur usually uses statistical-based methods in canonical segmentation, which relies on the artificial extraction of features. In surface segmentation, the artificial feature extraction process is avoided by using the neural network. However, to date, no model can provide both segmentation results in Uyghur without adding features. In addition, morphological segmentation is usually regarded as a sequence annotation task, so label imbalance easily occurs in datasets. Given the above situation, this paper proposes an improved labelling scheme that joins morphological boundary labels and voice harmony labels for the two kinds of segmentation simultaneously. Then, a convolution network and attention mechanism are added to capture local and global features, respectively. Finally, morphological segmentation is regarded as a sequence labeling task of character sequences. Due to the problem of label proportion imbalance and noise in the dataset, a focal loss function with label smoothing is used. The experimental results show that the F1 values of canonical segmentation and surface segmentation achieve the best results.http://dx.doi.org/10.1080/09540091.2022.2134843attention mechanismlabel imbalancemorphological segmentationtagging scheme |
| spellingShingle | Gulinigeer Abudouwaili Kahaerjing Abiderexiti Yunfei Shen Aishan Wumaier Research on the Uyghur morphological segmentation model with an attention mechanism Connection Science attention mechanism label imbalance morphological segmentation tagging scheme |
| title | Research on the Uyghur morphological segmentation model with an attention mechanism |
| title_full | Research on the Uyghur morphological segmentation model with an attention mechanism |
| title_fullStr | Research on the Uyghur morphological segmentation model with an attention mechanism |
| title_full_unstemmed | Research on the Uyghur morphological segmentation model with an attention mechanism |
| title_short | Research on the Uyghur morphological segmentation model with an attention mechanism |
| title_sort | research on the uyghur morphological segmentation model with an attention mechanism |
| topic | attention mechanism label imbalance morphological segmentation tagging scheme |
| url | http://dx.doi.org/10.1080/09540091.2022.2134843 |
| work_keys_str_mv | AT gulinigeerabudouwaili researchontheuyghurmorphologicalsegmentationmodelwithanattentionmechanism AT kahaerjingabiderexiti researchontheuyghurmorphologicalsegmentationmodelwithanattentionmechanism AT yunfeishen researchontheuyghurmorphologicalsegmentationmodelwithanattentionmechanism AT aishanwumaier researchontheuyghurmorphologicalsegmentationmodelwithanattentionmechanism |