cleaned_PhoMT

Dataset Summary

cleaned_PhoMT is a cleaned and quality-filtered subset of the PhoMT Vietnamese-English parallel corpus. The dataset is designed for machine translation research, particularly for training and evaluating Vietnamese ↔ English translation systems.

Starting from the original PhoMT corpus, we applied multiple data cleaning and quality control procedures to remove noisy, misaligned, and low-quality sentence pairs. The resulting dataset contains approximately 300,000 high-quality Vietnamese-English parallel sentence pairs.

Languages

• Vietnamese (vi)
• English (en)

Tasks

• Machine Translation

---

Data Cleaning Pipeline

The original PhoMT corpus was processed using a series of filtering steps to improve translation quality and sentence alignment.

1. Language Identification Filtering

Sentence pairs with incorrect language assignments were removed.

2. Length-Based Filtering

Very short and very long sentence pairs often introduce noise into machine translation training.

We filtered samples based on:

• Source sentence length.
• Target sentence length.
• Length ratio between source and target sentences.

Sentence pairs outside predefined thresholds were discarded.

3. Alignment Quality Filtering

To improve translation consistency, we applied alignment verification using:

Length Alignment

Source and target lengths were compared to identify abnormal translation pairs with large length discrepancies.

Semantic Alignment

Semantic similarity between source and target sentences was evaluated using multilingual sentence representations. Pairs exhibiting weak semantic correspondence were removed.

---

Limitations

Although extensive filtering was applied, some noisy or imperfect translation pairs may still remain.

---

Source Dataset

If you find my dataset useful, please cite to the original dataset, the PhoMT corpus:

PhoMT: A High-Quality and Large-Scale Benchmark Dataset for Vietnamese-English Machine Translation

Citation

@inproceedings{doan-etal-2021-phomt,
    title = {PhoMT: A High-Quality and Large-Scale Benchmark Dataset for Vietnamese-English Machine Translation},
    author = {Doan, Long and
      Nguyen, Linh The and
      Tran, Nguyen Luong and
      Hoang, Thai and
      Nguyen, Dat Quoc},
    editor = {Moens, Marie-Francine and
      Huang, Xuanjing and
      Specia, Lucia and
      Yih, Scott Wen-tau},
    booktitle = {Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing},
    month = nov,
    year = {2021},
    address = {Online and Punta Cana, Dominican Republic},
    publisher = {Association for Computational Linguistics},
    url = {https://aclanthology.org/2021.emnlp-main.369/},
    doi = {10.18653/v1/2021.emnlp-main.369},
    pages = {4495--4503}
}

---

License

Please follow the license and usage terms of the original PhoMT dataset. Users are responsible for ensuring compliance with the original dataset's licensing conditions.