Hi, I wonder whether the opennlp parser has certain connection to charniak parser and bikel parser? for example in terms of the algorithm, notations used etc?
I am using opennlp in an application, and now there is a need for using syntactic parser; I learnt from other literature that Charniak's and Bikel's parsers are well known in the field of bioinformatics, so I am wondering if i can use their parsers together with opennlp? Or maybe opennlp has certain relation with their parsers? Also, is there a general evaluation of the opennlp parser, compared to the others?
Any suggestions are much appreciated!
Thanks alot!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
There is a standard evaluation method, called parse-eval, which is used in the paper. OpenNlp results are comparable to those reported in the paper when trained on the same training data.
I'm not sure of the liciencing of the other parsers. Dan Bikel's parser used to be research only, not sure about Charniak's. Also, these parsers are only trained on Wall Street Journal text. The models distributed with OpenNlp include some additional data to make it more suitable for a wider variety of text. OpenNlp also has considerably better run-time performance.
Hope this helps...Tom
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hi, I wonder whether the opennlp parser has certain connection to charniak parser and bikel parser? for example in terms of the algorithm, notations used etc?
I am using opennlp in an application, and now there is a need for using syntactic parser; I learnt from other literature that Charniak's and Bikel's parsers are well known in the field of bioinformatics, so I am wondering if i can use their parsers together with opennlp? Or maybe opennlp has certain relation with their parsers? Also, is there a general evaluation of the opennlp parser, compared to the others?
Any suggestions are much appreciated!
Thanks alot!
Hi,
This parser is based on the algorithms described in Adwait Ratnaparkhi's thesis. A short version of that work is at: http://acl.ldc.upenn.edu/W/W97/W97-0301.pdf
There is a standard evaluation method, called parse-eval, which is used in the paper. OpenNlp results are comparable to those reported in the paper when trained on the same training data.
I'm not sure of the liciencing of the other parsers. Dan Bikel's parser used to be research only, not sure about Charniak's. Also, these parsers are only trained on Wall Street Journal text. The models distributed with OpenNlp include some additional data to make it more suitable for a wider variety of text. OpenNlp also has considerably better run-time performance.
Hope this helps...Tom
Thanks so much!