Charniak how-to

Charniak is one of the best morfosyntactic parsers for English since 2005.

The distribution is precompiled for 32-bits. For 64bits RECOMPILE: en parser05May26fixed

COMPILATION ————————————————-

If 64-bit: You may need to install lib32 lib32std … rm -rf *.o

Both 32 and 64-bit: Since Ubuntu 9.10 you may also need to manually download and install (dpkg -i) getlibs-all.deb and libstdc++5_3.3.6-17ubuntu1_i386.deb.

make

If you have compilation errors see BchartSm.C (insert #include “GotIter.h”) and AnsStrCounts.C (insert #include “AnswerTree.h”   and   #include “math.h”).

make parseIt

——————————————————————–

REMEMBER: chmod a+x CharniakServer/parser05May26fixed/parseIt

cd charniakserver

vim charniak-server.pl –> update $charniakDir with current path

To execute the server: sudo ./ejecuta_charniak.sh  (parseIt process)

To kill the server: ./kill_charniak.sh

To test it: execute prueba_charniak.sh (or echo “…” | ./charniak-client.pl) or telnet localhost 4449 –> input sentence + enter

Charniak does not split sentences. It needs one sentence per line. (HOWEVER SRL tool splits sentences perfectly)

Works in UTF-8.