What if u try to use Blue system which is used in evaluation MT systems
Good luck Bayan
Sent from my iPhone
> On Oct 27, 2014, at 8:14 PM, John F Sowa <sowa at bestweb.net> wrote:
> Since nobody responded to this question in the past two weeks,
> I thought I might make some suggestions.
>> On 10/10/2014 6:15 AM, Emad Mohamed wrote:
>> Do you know of any work on the automatic evaluation of human translation?
>> Typical scenario from my school:
>> 200+ students per class in a translation studies course and one
>> instructor. If the students can upload their translations and be given
>> automatic feedback, this will greatly facilitate the process.
> One possibility is to adapt software used for evaluating answers
> to essay questions. For example,
> 1. An expert on the subject writes a "gold standard" essay on
> the same topic. The gold essay covers the points that the
> students should make for an A+ score.
> 2. The software checks whether the expected points (in related
> terminology) are mentioned in students' essays, and it uses
> various methods for evaluating spelling, grammar, style, etc.
> In addition to the methods for evaluating essays, MT methods for
> aligning sentences and evaluating possible word and phrase choices
> could be adapted to this task.
> The instructor could also list erroneous choices that should be
> avoided. When running the software, the instructor could check
> some evaluations, modify the list of preferred and erroneous
> choices, and rerun all evaluations with the updated list.
> Has anybody implemented anything along these lines? Or anything
> else that might be useful for developing such software?
> Software designed for evaluating human translations could also
> be use to evaluate MT systems -- and vice versa.
> UNSUBSCRIBE from this page: http://mailman.uib.no/options/corpora
> Corpora mailing list
> Corpora at uib.no