First, ensure that you have downloaded C&C parser
and EasyCCG parser and wrote their installation locations
in the files en/candc_location.txt and en/easyccg_location.txt, respectively.
Second, you need to download the SICK dataset and VerbOcean by running the following script:
./en/download_dependencies.shYou can evaluate the end-to-end system performance of a certain list of semantic templates on the test split of SICK by doing:
./en/eacl2017exp.sh 10 test en/semantic_templates_en_event.yamlThis script will coordinate the tokenization, syntactic parsing (with C&C and
EasyCCG), semantic parsing and theorem proving (with Coq) using 10 processes.
Syntactic and semantic parsing results will be written in parsed directory.
Entailment judgements and an HTML graphical representation of semantic
composition (and constructed theorem) will be written in results directory.
You can see a summary of performance by doing:
cat results/score.txtand you should see something similar to this:
Correct parsing: 0.9748 (4803/4927)
Accuracy: 0.8313 (4096/4927)
Recall: 0.6265
Precision: 0.9688
F1 score: 0.7608
Gold_correct_total: 2134
System_answer_total: 1380
System_correct_total: 1337
----------------------------------------------------------------
system
| | yes | no | unknown | error | total
----------------------------------------------------------------
| yes| 838 | 6 | 565 | 5 | 1414
gold | no| 3 | 499 | 176 | 42 | 720
| unknown| 22 | 12 | 2682 | 77 | 2793
| total| 863 | 517 | 3423 | 124 | 4927
----------------------------------------------------------------
If you want to see the results (syntactic/semantic parses, entailment judgements and HTML
visualizations) but do not wish to run the software, you can uncompress the file
en/sick_intermediate_results.tgz by doing:
tar xvzf en/sick_intermediate_results.tgzwhich will create the plain/, parsed/ and results/ directories.