Comparative Assessment with Annotation on Paragraphs (CAP) Job execution

The CAP job is based on the Comparative Assessment with added support for dynamic annotation on Paragraph context.

The CAP operates on whole paragraphs. The first step is paragraph segmentation including segment alignment. The evaluator then scores and annotates MT output from two different engines.

The dynamic annotation fields are displayed on the bottom of each source and translation cards. When the annotation field is marked with asterisk, it must be fullfilled before confirming the segment evaluation.

Tie Breaker

If the Tie Breaker is enabled in Job configuration, the evaluator is asked to select the better translation when both translation have the same scores.