Anyone who has the Jacy book could kindly give me some coverage figure from it? What does it report, over what kind of corpus? Many thanks.
I think these are the tables you are looking for. The prose on the previous page says how the “kinou” sets are a machine translation test suite and how of the two “natural” corpora, the tanaka corpus was used more as a development corpus while the haikingu (“hiking”) corpus is unseen and thus a better measure of the grammar’s ability to generalize.
1 Like
Thanks from me too!