NLP-ps3

Natural Language Processing - Problem Set 3

installing dependencies

All dependencies should be bundled with an Anaconda install, if any are missing they can be installed using pip and the requirements.txt file. Run pip install -r requirements.txt to install missing dependencies on your system.

How to Run

Assuming you are in the root directory of the project, the program can be run from the command line with python3 main.py.

Usage

Help output is as follows

$ python3 main.py 
usage: main.py [-h] [--train TRAINING_DATA] [--test TEST_DATA] [-m]

Train or test classifiers.

optional arguments:
  -h, --help            show this help message and exit
  --train TRAINING_DATA
                        Train the classifiers with the given training data
                        file.
  --test TEST_DATA      Test the classifiers with the given training data
                        file. Prints out raw prediction data by default
  -m                    Print basic test metrics instead of raw predictions.

Project Structure

• main.py - the main program, run this from the commandline
• /trained_models - this directory will contained the serialized "pickled" trained models that get generated in "--train" mode. These are what will be used when using the "--test" mode of the progam.
• /classifiers - this directory contains separate code for each of the 3 classifiers
  • init.py - this file tells python to treat this directory as a moduke, allowing us to import the classifiers from within the main.py file.
  • eventType.py - contains the code for the multi-class event type classifier.
  • genre.py - contains the code for the binary genre classifier.
  • polarity.py - contains the code for the 3-way polarity classifier