| Title |
Programavimo uždavinių sudėtingumo automatinio įvertinimo tyrimas / |
| Translation of Title |
Research on automatic difficulty estimation of programming problems. |
| Authors |
Skarbalius, Artūras |
| Full Text |
|
| Pages |
48 |
| Keywords [eng] |
programming problem ; difficulty evaluation ; natural language processing ; convolutional neural network ; deep learning |
| Abstract [eng] |
This document details an attempt to use machine learning for automatic evaluation of the difficulty of programming problems or exercises. Typically, the problems consist of both text description and accompanying figures. As such, evaluation of various possible models for both text and images is required. The goal of this project is to determine whether this sort of machine learning can reliably evaluate the difficulty of programming exercises. The work involves collecting and creating a suitable dataset, analyzing the evaluation based on the text and the image data separately, and a method of combining the two methods is then attempted. The results of this investigation are reported near the end of the document, alongside with potential theoretical methods to improve the system further. |
| Dissertation Institution |
Kauno technologijos universitetas. |
| Type |
Master thesis |
| Language |
Lithuanian |
| Publication date |
2021 |
