Parallel Corpus Sentence Alignment Scoring Dataset for Low-Resource Languages 低资源语言平行语料句对齐评分数据集

Based on the neural network-based unsupervised sentence embedding method NeuroAlign, parallel sentence pairs are embedded into the same vector space, and alignment scores for given candidate sentence pairs in the parallel corpus are calculated. Then, low-scoring sentence pairs are filtered out based on these scores, resulting in high-quality bilingual parallel corpora for low-resource languages. Based on the neural network-based unsupervised sentence embedding method NeuroAlign, parallel sentence pairs are embedded into the same vector space, and alignment scores for given candidate sentence pairs in the parallel corpus are calculated. Then, low-scoring sentence pairs are filtered out based on these scores, resulting in high-quality bilingual parallel corpora for low-resource languages.