Experimental Settings for "Towards Joint Morphological Analysis and Dependency Parsing for Turkish", DepLing 2013



We used the detached version of the Turkish treebank.  The original data has both coarse- and fine-grained POS tags. The parsers used in experiments expect the CoNLL09 format where there is only one POS field. We used only the fine-grained tags in the experiments. 

Parser Settings

The graph-based parser:

java -Xmx24g  -cp anna-660.jar is2.parser52LX2CS.Parser -train trainFile  -model modelFile -test inFile -out parsedFile -eval goldFile

The joint parser:

java -Xmx24g  -cp anna-660.jar is2.transitionR5ysp7b.Parser -train trainFile -test inFile -out parsedFile  -model modelFile -eval goldFile  -i 25  -hsize 400000001 -cores 12  -beam 80 -2nd abcd -3rd abc -1st a -tsize 2 -tnumber 10 -ti 10  -x train:test -thsize 90000001 -tthreshold 0.3 -tx 3  -decoder nv -half 4  -tt 25

The parameter descriptions for the joint parser can be found here.


We used an IG-based evaluator from Gülşen Eryiğit. The files to be compared are in CoNLL06 format. 

java -cp . Parsing_RawdataConll_Evaluator goldFile inFile

For the upper bounds with 100% MWE, we set the parameter


within the code.