MUSE

该数据集包含了多语言对的的双语词典，用于训练和评估跨语言词嵌入（CLWE）。此外，该数据集被用于测试不同的CLWE方法，并通过迭代归一化评估翻译准确性的提升。在规模上，训练集包含5000个源语言词汇，并以1500个源语言词汇的准确度进行报告。该数据集的任务是词典诱导基准测试以及词汇翻译准确性的评估。

This dataset contains bilingual dictionaries for multiple language pairs, which are utilized for training and evaluating Cross-Lingual Word Embeddings (CLWE). Furthermore, this dataset is employed to assess diverse CLWE approaches and measure the improvement in translation accuracy through iterative normalization. Regarding its scale, the training set includes 5000 source language vocabulary items, and the evaluation accuracy is reported based on 1500 source language vocabulary items. The tasks of this dataset include dictionary induction benchmarking and lexical translation accuracy evaluation.