Categorizing Short Texts: A Comparison of Network Theory and Neural Network Approaches

This project aims to find effective ways of classifying collections of short texts such as Twitter tweets or flashcards into meaningful categories. Specifically, my aim is to classify collections of flashcards in a spaced repetition system, which is a program used to manage large collections of flashcards, widely used to learn and review bodies of knowledge such as foreign languages, medical specializations, or fields of science. Generally they store a database of question-answer pairs created by the user. When the user learns or reviews a card, they are prompted to rate the difficulty of remembering the answer. This data is then used to compute the optimal time to next review the card, according to statistical laws of memory first derived in the PhD. thesis of researcher Piotr Wozniak (TODO cite). Such careful scheduling of reviews optimizes the retention of memories while minimizing the work required.