Pronomial Coreferencing

This is an NLP project for the English language which aims to resolve the pronouns present in a piece of text to reflect the noun that they refer to.

Example: Aaron telephoned Jake to tell him that he lost the laptop. Aaron telephoned Jake to tell Jake that Aaron lost the laptop.

Prerequisites

1. Python 3.3+ interpreter
2. Python packages that need to be installed are mentioned in requirements.txt

Getting Started

Here is a quick guide to the various files to help you navigate and use the repository.

gender_classify.py

A Machine Learning script that trains a model to classify people names into male or female.

main.py

Core computation is done in this script. Each function in the script does the following:

• preprocess() - Contructs a dependency parse and extracts sentence features
• resolution() - Maps each pronoun to its noun
• build_output() - Constructs the output string that is to be returned
• main() - Accepts an input string and returns the resolved string

run.py

The script that must be run to access main.py. It provides two alternative functions to take input from the user:

• file() - Provide input as a text file, input.txt
• keyboard() - Manually type text as input

Call one of the above functions that you need to use in run.py as shown:

if __name__ == '__main__':
    keyboard()

data folder

• names.csv - Dataset of people names and the corresponding gender of the person. Used to train the classifier in gender_classify.py
• input.txt - The text file where you can save your input text which needs to be resolved for pronouns