English-Malay Cross-Lingual Emotion Detection In Tweets Using Word Embedding Alignment
The three-phase methodology to address the goals of this study included the construction of the English-Malay cross-lingual word embedding using word embedding alignment, enrichment of the cross-lingual word embedding with sentiment information, and pre-training of the hierarchical attention model s...
Saved in:
| 主要作者: | |
|---|---|
| 格式: | Thesis |
| 語言: | English |
| 出版: |
2023
|
| 主題: | |
| 在線閱讀: | http://eprints.usm.my/60433/1/Pages%20from%20LIM%20YING%20HAO%20-%20TESIS.pdf |
| 標簽: |
添加標簽
沒有標簽, 成為第一個標記此記錄!
|
| 總結: | The three-phase methodology to address the goals of this study included the construction of the English-Malay cross-lingual word embedding using word embedding alignment, enrichment of the cross-lingual word embedding with sentiment information, and pre-training of the hierarchical attention model solely on English tweets. We evaluated our model in two scenarios: zero-shot learning and few-shot learning on 4176 Malay tweets annotated with emotion. We also examined the optimal number of Malay tweets required to finetune the model and the effect of finetuning different layers in our model. |
|---|