Accompanying dataset for: A Monte Carlo Method for Metamorphic Testing of Machine Translation Services

This dataset includes enhanced analysis of the machine translation data. The original dataset has been reported in [1], where white spaces were used to separate words in different languages. This is however not the best method of analyzing some Asian languages such as the Chinese language. In the present analysis, we used a character-based approach to separating the Chinese and Japanese results, hence obtaining a different set of BLEU and Cosine Similarity scores. These new scores are given in the present dataset.

[1] Daniel Pesu, Zhi Quan Zhou, Jingfeng Zhen, & Dave Towey. (2018). Accompanying dataset for: A Monte Carlo Method for Metamorphic Testing of Machine Translation Services (Version 1.0) [Data set]. Zenodo. http://doi.org/10.5281/zenodo.1194560