Word Problem Solver

Prerequisites:

• NLTK 3.1
• ScikitLearn 0.16.1
• NumPy
• Install MaxEnt 'megam'

Data

Data is provided in the data/ folder. Follow the format given in data/questions.json

• Index
• Wordproblem
• Equation
• Solution

Run

Initially, we convert all equations into generic equations (equation templates) We then divide the dataset into two halves - one is for training, the other is for testing.

Execute the following command: python Parser.py

You can see the accuracy for each of the classifier in both NLTK and Sklearn libraries. The templates for each wordproblem will be available in data/templates.txt

References

• http://textminingonline.com/dive-into-nltk-part-viii-using-external-maximum-entropy-modeling-libraries-for-text-classification
• http://www.nltk.org/
• http://scikit-learn.org/stable/