Machine Translation Evaluation Dataset for Amharic

<strong>Machine Translation Evaluation Dataset for Amharic </strong> The dataset contains source sentences in Amharic and English and their corresponding reference translations that were collected using crowd sourcing. These ground-truth sentences are from across different domains such as news headlines, social media, Wikipedia and everyday conversation. <br> <strong>Metadata of files in the dataset</strong> amen.tsv<br> - Domain: news | wiki | twitter | convo<br> - Source Sentence: Amharic sentence<br> - Reference Translation: English translation<br> - Google Translate: output of Google Translate<br> - Yandex Translate: output of Yandex Translate <br> enam.tsv<br> - Domain: news | wiki | twitter | convo<br> - Source Sentence: English sentence<br> - Reference Translation: Amharic translation<br> - Google Translate: output of Google Translate<br> - Yandex Translate: output of Yandex Translate <br> <strong>Amharic source and reference translations across domains:</strong> <em>News: </em>These are news headlines from Ethiopian news websites.<br> <em>Wikipedia: </em>A random sample of sentences from the Amharic Wikipedia.<br> <em>Twitter: </em>Amharic Twitter posts on consumer products.<br> <em>Conversational: </em>Everyday conversational expressions from Amharic native speakers. <strong>English source and reference translations across domains:</strong> <em>News: </em>These are news headlines from Wikipedia current events portal.<br> <em>Wikipedia: </em>A random sample of sentences from the English Wikipedia.<br> <em>Twitter: </em>English Twitter posts on global events from Wikipedia current events portal.<br> <em>Conversational: </em>Everyday conversational expressions from English native speakers. <strong>Evaluation of two systems that provide Amharic translation</strong> The dataset also contains evaluation of two commercial systems: [Google<br> Translate](https://translate.google.com/) and [Yandex<br> Translate](https://translate.yandex.com/). Both systems provide free APIs that<br> users can sign up and get access keys to. The translations for Amharic to English were generated on 14th<br> February 2020. The translations for English to Amharic were generated on 30th March 2020.