ADRT BOXER

What is ADRT Boxer?

Turkish ADRT Boxer is a prototype DRT boxer that analyzes input sentences of Turkish in temporal domain by using the ADRT framework. Turkish ADRT Boxer is implemented as a term project for the fullfilment of COGS 527 Dynamics Semantics course requirements.

Download

Source code for the Turkish ADRT Boxer can be downloaded from here.

Installation

Turkish ADRT Boxer is implemented with Python 2.7 and NLTK. Therefore, before compiling the boxer Python 2.7 and NLTK must be installed on your system. This final release have been tested on Ubuntu 12.04 LTS, hence we also recommend you to use Ubuntu while working on the final release. Ubuntu 12.04 can be downloaded from this link. You might encounter problems while compiling this release in other OS. We do not guarantee that it will work on other OS without a hitch. If you are going to run release on Ubuntu, you don't need to install Python from scratch since it comes bundled with the OS. Yet, you need to follow the instructions from this page to install NLTK and all other necessary dependencies.

Compilation

In order to compile the boxer extract the source code from turkish_adrt_boxer.zip. Start up a shell and traverse into the directory that contains the source code. If you type in "python boxer.py" to the shell while your in the directory that contains the source code boxer will analyze the input sentence and draw a ADRT box representation of the input sentence just for the temporal domain. To change the input sentence you must edit the line "trees = parser.nbest_parse('HERE COMES THE INPUT SENTENCE'.split())" in the boxer.py file. All lexical items that can be used in the input sentence and context free grammar rules that define the syntax can be found in file named grammar.fcfg.

Project Team