The MaltBlender webpage

MaltBlender is an tool for merging the output of several dependency parsers, using the Chu-Liu/Edmonds directed MST algorithm (Sagae and Lavie 2006). This software was used by the wining team of the CoNLL-XI Shared Task (Hall et al. 2007).

MaltBlender requires Java JRE 1.5 or later, Perl v5.8.1 or later, and the evaluation script eval07.pl from CoNLL-XI Shared Task 2007.

Run MaltBlender: java -DPERL=<perl_path> -DEVALUATOR=<eval07.pl_path> -jar MaltBlender.jar -w 4 -o out.conll -l 4 -h <held-out_gold_file> -H <held-out_parsed_files> -F <parsed file>

-DPREL = path to perl
-DEVALUATOR = path to eval07.pl
-h = the gold-standard file for the held-out data, which is used for estimating the weights together with the parsed files after -H
-H = the parsed file for the held-out data; separate the files wth a white space
-F = parsed file to be combined (make sure that the number of files after -H and -F is the same. Note that the ith file after -H is used for estimating the weight of the ith file after -F)

Note that the two first arguments, -DPREL and -DEVALUATOR, have to be located before "-jar", since they are arguments to the java interpreter, not MaltBlender.

Download MaltBlender

References

Sagae, K. and Lavie, A. 2006. Parser combination by reparsing. Proceedings of the 2006 Human Language Technology Conference of the North American Chapter of the Association for Computational Linguistics - short papers (HLT-NAACL'06). New York, NY.
[PDF] J. Hall, J. Nilsson, J. Nivre, G. Eryigit, B. Megyesi, M. Nilsson, M. Saers. (2007) Single Malt or Blended? A Study in Multilingual Parser Optimization. In Proceedings of the Conference on Empirical Methods in Natural Language Processing and Conference on Computational Natural Language Learning (EMNLP-CoNLL), Prauge, Czech Republic, pp. 933-939