Arabic standard parsing experiments (test set, sentences ≤ 40 words). SplitPCFG is the same grammar used in the Stanford parser, but without the dependency model. FactLex uses basic POS tags predicted by the parser and morphological analyses from MADA. FactLex* uses gold morphological analyses. Berkeley and DP-TSG results are the average of three independent runs.
Arabic . | Leaf Ancestor . | Evalb . | ||||
---|---|---|---|---|---|---|
. | Sent. . | Corpus . | LP . | LR . | F1 . | EX% . |
PAPCFG | 0.777 | 0.745 | 69.5 | 64.6 | 66.9 | 12.9 |
SplitPCFG | 0.821 | 0.797 | 75.6 | 73.4 | 74.5 | 17.8 |
Berkeley | 0.865 | 0.853 | 83.3 | 82.7 | 83.0 | 24.0 |
DP-TSG | 0.822 | 0.800 | 75.5 | 75.4 | 75.4 | 17.7 |
Stanford | 0.851 | 0.835 | 81.3 | 80.7 | 81.0 | 23.5 |
Stanford+FactLex | 0.849 | 0.835 | 81.2 | 80.8 | 81.0 | 22.8 |
Stanford+FactLex* | 0.852 | 0.837 | 81.8 | 81.3 | 81.5 | 24.0 |
Arabic . | Leaf Ancestor . | Evalb . | ||||
---|---|---|---|---|---|---|
. | Sent. . | Corpus . | LP . | LR . | F1 . | EX% . |
PAPCFG | 0.777 | 0.745 | 69.5 | 64.6 | 66.9 | 12.9 |
SplitPCFG | 0.821 | 0.797 | 75.6 | 73.4 | 74.5 | 17.8 |
Berkeley | 0.865 | 0.853 | 83.3 | 82.7 | 83.0 | 24.0 |
DP-TSG | 0.822 | 0.800 | 75.5 | 75.4 | 75.4 | 17.7 |
Stanford | 0.851 | 0.835 | 81.3 | 80.7 | 81.0 | 23.5 |
Stanford+FactLex | 0.849 | 0.835 | 81.2 | 80.8 | 81.0 | 22.8 |
Stanford+FactLex* | 0.852 | 0.837 | 81.8 | 81.3 | 81.5 | 24.0 |