kelliemwebster Goto Github PK

gap-coreference

natgenmt

This dataset is intended as an evaluation benchmark for gender issues in Machine Translation. We consider the challenges in modeling and handling gendered language in the context of machine translation and extend over previous work that identifies issues using synthetic examples. We focus on the class of issues which surface when a neutral reference to a person is translated to a gendered form. For this class of examples, the MT task requires a system to produce a single translation without source cues, thus exposing a model's preferred gender for the reference form. We include English source sentences, and four target gendered languages across three language families (French, German, Spanish, and Russian). The examples included in the dataset expose where MT encodings are gendered, finding new issues not covered in previous manual approaches.

noun-verb

This dataset contains naturally-occurring English sentences that feature non-trivial noun-verb ambiguity.