Math word problem to the equation

Industry: Edutech Platform

Edutech Platform

Overview

Natural Language processing custom built deep learning algorithms to generate mathematical equations from Math word problem statements without human interference. During this pandemic we saw a sudden rise of internet education & need to have tutors for children when they have doubts. So a startup allowed us to build an app where Students or parents can find out equations for the problem they enter within the app itself.

Technology

  • Deep Learning
  • NLP
  • Machine Learning

Language

  • Python

Key Technical Challenges:

  • Less amount of data to train our deep learning model. Data available for only uni variable equations.
  • Build a custom neural network which can understand data and generate polynomial equations.
  • Had to imply some research into this for machine translation problems.

Business + Technical Points:

  • Build a custom deep learning system to generate linear equations from natural language word problem statements.
  • Systems down the line should be able to fit more than 1000 requests per second
  • Easily modifiable code for his team to understand & work with.

Result:

  • We are able to build a system to generate mathematical equations of word problems.