Hello,
Using tccg I can parse and display the semantics of a single example sentence. :2xml also saves this single example to xml. Is it possible to choose a sentence LIST (e.g. testbed) as input and save all parsed semantics to file? How?
Thank you!
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
At present it seems to be missing a command-line script in $OPENCCG_HOME/bin/, but it would be easy to add a ccg-parse (and ccg-parse.bat) script to do so. If you create these, you could even issue a pull request on github to add them to the current dev version.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Right, it saves the LF output of each parse to a testbed file.
I've added ccg-parse scripts to the master branch on github, but I'm reminded that it's currently set up in a way that requires a supertagging and parsing model. As such, it's easier to use the ant-based script for parsing novel text that also does necessary pre-processing, which is described in docs/ccgbank-README in the section "Using the pre-built English models". Otherwise, if you're not using a statistical model, then it will perhaps make sense to write a simpler version of the program in src/opennlp/ccg/Parse.java.
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Hello,
Using tccg I can parse and display the semantics of a single example sentence. :2xml also saves this single example to xml. Is it possible to choose a sentence LIST (e.g. testbed) as input and save all parsed semantics to file? How?
Thank you!
Hi
There is a program to parse a file with one sentence per line and output a testbed file: https://github.com/OpenCCG/openccg/blob/master/src/opennlp/ccg/Parse.java
At present it seems to be missing a command-line script in $OPENCCG_HOME/bin/, but it would be easy to add a ccg-parse (and ccg-parse.bat) script to do so. If you create these, you could even issue a pull request on github to add them to the current dev version.
Hi Michael, thank you!
With "output a testbed" you mean that I could save the semantic output of every input line to file?
Right, it saves the LF output of each parse to a testbed file.
I've added ccg-parse scripts to the master branch on github, but I'm reminded that it's currently set up in a way that requires a supertagging and parsing model. As such, it's easier to use the ant-based script for parsing novel text that also does necessary pre-processing, which is described in docs/ccgbank-README in the section "Using the pre-built English models". Otherwise, if you're not using a statistical model, then it will perhaps make sense to write a simpler version of the program in src/opennlp/ccg/Parse.java.