FI MU, autumn 2011, IA161 syntactic formalisms for natural language parsing
IA161 Final homework
Task : due on January 6th
2012
Deadline: January 13th
2012, 05:00 p.m
1. Installation and evaluation of the parsers
Download all the materials (parsers, instruction, data) from the following web page:
https://is.muni.cz/el/1433/podzim2011/IA161/30264269/
For evaluating the RASP, CCG Parser and Standford Parser, the same input sentences will be used.
Download data.tar.gz which include the input sentences to be parsed and unpack it. The input
sentences are extracted from the NYT.
1.1 RASP
Download the RASP rasp3os.tar.gz and unpack it to get rasp3os directory.
$ tar xzvf rasp3os.tar.gz
$ cd rasp3os/scripts
Run the RASP with the following command:
$./rasp.sh ­m <test.txt> result.txt
1.2 CCG Parser
Download the CCG Parser candc­linux­1.00.tgz and models­1.02.tbz2 in two steps.
First, unpack candc­linux­1.00.tgz to get candc-1.00 directory which include the software tool,
then unpack models­1.02.tbz2 to get models directory which include the trained models form the
Penn Treebanks under the candc-1.00 directory.
$ tar xzvf candc­linux­1.00.tgz
$ cd candc­1.00
$ tar xzvf ../models­1.02.tbz2
Run the CCG Parser with the following command:
$ candc­1.00/bin/candc ­­models=candc­1.00/models \­­input data/test ­­output 
data/test.candc
1.3 Standford Parser
Download the Standford Parser stanford­parser­2011­09­14.tgz  and unpack it to get stanfordparser-2011-09-14
directory.
Run the Standford Parser with the following command:
$ ./lexparser.sh data/test.txt
FI MU, autumn 2011, IA161 syntactic formalisms for natural language parsing
2. Analysis and comment on the obtained results
a) Analyze the quality of parsing outputs. Compare and comment at least 5 phenomena contained in
the input sentences, for example, parentheses, punctuation, reflexive verb, relative clause, modal verb
followed by a verb in the infinitive form, prepositional clausal modifier, temporal modifier, multi-word
expression, etc. You can evaluate the parsing results manually by using your linguistic intuition.
b) Report the errors that you found during the analysis of the outputs.
c) Compare the parsing method (techniques, theories) used in the three parsers. Remind you the
technical and theoretical backgrounds presented in the lectures of this course.
Attention!
a) Please attach your outputs (screen-shots) on the final file.
b) Please send one compressed file which include all the documents: copies of outputs, analysis of the
results,etc to qkang@fi.muni.cz.