Competition 2
English
Table: The obtained average precision (AP%)
in the information retrieval task for the submitted segmentations
in English. Methods without authors are reference methods by
the organizers. The types are: S = semi-supervised algorithm, P =
unsupervised algorithm with supervised parameter tuning, U = fully
unsupervised algorithm. Indexing is performed using Okapi (BM25) weighting
for all morphemes except the most common ones (stoplist) with
frequency higher than 150,000.
AUTHOR |
METHOD |
AP% |
TYPE |
- |
snowball porter |
0.4092 |
- |
- |
TWOL first |
0.4020 |
- |
- |
TWOL all |
0.3994 |
- |
- |
Morfessor Baseline |
0.3835 |
U |
Lignos |
Base Inference |
0.3832 |
U |
Kohonen |
Morfessor S+W+L |
0.3820 |
S |
Lignos |
Aggressive Compounding |
0.3784 |
U |
- |
grammatical first |
0.3777 |
- |
Kohonen |
Morfessor S+W |
0.3776 |
S |
Lignos |
Iterative Compounding |
0.3776 |
U |
- |
Morfessor CatMAP |
0.3754 |
U |
Kohonen |
Morfessor U+W |
0.3690 |
P |
Nicolas |
MorphAcq |
0.3680 |
U |
- |
grammatical all |
0.3567 |
- |
Spiegler |
Promodes-E |
0.3379 |
P |
- |
dummy |
0.3304 |
- |
Spiegler |
Promodes |
0.3286 |
P |
Spiegler |
DEAP PROB-NOCAT |
0.3141 |
S |
Spiegler |
DEAP MDL-NOCAT |
0.2916 |
S |
Spiegler |
DEAP MDL-CAT |
0.2409 |
S |
Golenia |
MAGIP |
0.2257 |
S |
Spiegler |
DEAP PROB-CAT |
0.2186 |
S |
Spiegler |
Promodes-H |
0.2034 |
P |
Finnish
Table: The obtained average precision (AP%)
in the information retrieval task for the submitted segmentations
in Finnish. Methods without authors are reference methods by
the organizers. The types are: S = semi-supervised algorithm, P =
unsupervised algorithm with supervised parameter tuning, U = fully
unsupervised algorithm. Indexing is performed using Okapi (BM25) weighting
for all morphemes except the most common ones (stoplist) with corpus
frequency higher than 75,000.
AUTHOR |
METHOD |
AP% |
TYPE |
- |
TWOL first |
0.4973 |
- |
Lignos |
Aggressive Compounding |
0.4914 |
U |
- |
TWOL all |
0.4801 |
- |
- |
Morfessor CatMAP |
0.4754 |
U |
Kohonen |
Morfessor S+W |
0.4750 |
S |
Kohonen |
Morfessor S+W+L |
0.4465 |
S |
- |
grammatical first |
0.4317 |
- |
- |
snowball finnish |
0.4275 |
- |
- |
Morfessor Baseline |
0.4235 |
U |
Lignos |
Iterative Compounding |
0.4229 |
U |
Spiegler |
DEAP MDL-NOCAT |
0.4160 |
S |
Lignos |
Base Inference |
0.4151 |
U |
- |
grammatical all |
0.4093 |
- |
Kohonen |
Morfessor U+W |
0.4042 |
P |
Golenia |
MAGIP |
0.3871 |
S |
Spiegler |
DEAP MDL-CAT |
0.3725 |
S |
Spiegler |
Promodes |
0.3721 |
P |
- |
dummy |
0.3529 |
- |
Spiegler |
Promodes-E |
0.3213 |
P |
Spiegler |
DEAP PROB-NOCAT |
0.3174 |
S |
Spiegler |
Promodes-H |
0.3109 |
P |
Spiegler |
DEAP PROB-CAT |
0.2843 |
S |
German
Table: The obtained average precision (AP%)
in the information retrieval task for the submitted segmentations
in German. Methods without authors are reference methods by the
organizers. The types are: S = semi-supervised algorithm, P =
unsupervised algorithm with supervised parameter tuning, U = fully
unsupervised algorithm. Indexing is performed using Okapi (BM25) weighting for
all morphemes except the most common ones (stoplist) with frequency
higher than 150,000.
AUTHOR |
METHOD |
AP% |
TYPE |
- |
TWOL first |
0.4836 |
- |
- |
TWOL all |
0.4745 |
- |
Kohonen |
Morfessor U+W |
0.4666 |
P |
- |
Morfessor CatMAP |
0.4657 |
U |
Lignos |
Base Inference |
0.4434 |
U |
- |
Morfessor Baseline |
0.4391 |
U |
Lignos |
Aggressive Compounding |
0.4379 |
U |
Lignos |
Iterative Compounding |
0.4200 |
U |
Nicolas |
MorphAcq |
0.3862 |
U |
- |
snowball german |
0.3859 |
- |
- |
dummy |
0.3508 |
- |
- |
grammatical first |
0.3350 |
- |
- |
grammatical all |
0.3008 |
- |
Return to the result page
HOME
| RULES
| SCHEDULE
| DATASETS
| EVALUATION
| WORKSHOP
| FAQ
| CONTACT
Page maintained by webmaster at cis.hut.fi,
last updated Tuesday, 17-Aug-2010 16:36:25 EEST