Math Language Processing

My Penn State Schreyer undergraduate thesis work. Involves supplying attentional encoder decoder LSTMs and GRUs with elementary math problems and testing problem solving capabilities.

The ultimate application of doing so was to develop education assistive software tools. The output papers propose the application of these models for the automatic generation of multiple choice question distractors given a sequence of provided test questions.

Getting Started

This code was designed to be quickly and efficiently executable via command line to support rapid experimentation with model variables.

Using math_language_processing/main.py I would execute either a random forest, a GRU, or a now depricated LSTM on my schools on-prem server which I could SSH onto via my own personal laptop.

Style & Formatting

Code was written following PEP8 guidelines. Black was used to help enforce this.

• PEP8 guidelines
• Black code linter and formatter

Related Publications

Output publications from this project:

• https://honors.libraries.psu.edu/catalog/6933rob5372
• Bakes, Riley Owen, et al. “Educational Data Mining 2021.” Journal of Educational Data Mining, Math Question Solving and MCQ Distractor Generation with Attentional GRU Networks.

Paper which inspired the thesis:

• https://arxiv.org/abs/1904.01557