SimpliCT: A Digital Tutor Designed to Aid in the Education of Computational Thinking
Authors
Jensen, Jacob Guldhammer ; Svare, Jonas Jerup
Term
4. term
Education
Publication year
2024
Pages
14
Abstract
Computational thinking er en måde at gå til problemer på ved at bruge grundlæggende ideer fra datalogi, fx at dele opgaver op i trin, finde mønstre og lave simple regler. Vi mener, at det kan hjælpe folk med hverdagens udfordringer og styrke faglige kompetencer. I dette projekt undersøger vi, hvordan en digital tutor kan støtte læring i computational thinking. Vi udviklede en chatbot-tutor, SimpliCT, baseret på en GPT-sprogsmodel fra OpenAI, og finjusterede den med 68 træningseksempler. Derefter gennemførte vi et komparativt studie for at vurdere både brugeroplevelsen (hvordan det er at bruge systemet) og effektiviteten (hvad man lærer). Brugeroplevelsen var samlet set acceptabel, med tydeligt potentiale for forbedring. Vi kunne ikke drage sikre konklusioner om effektiviteten, men resultaterne tyder på, at nybegyndere kan opnå introducerende viden om computational thinking.
Computational thinking is a way of approaching problems using core ideas from computer science, such as breaking tasks into steps, spotting patterns, and designing simple rules. We believe it can help people tackle everyday challenges and strengthen academic skills. In this project, we explore how a digital tutor can support learning computational thinking. We built a chatbot tutor, SimpliCT, based on an OpenAI GPT language model and fine-tuned it with 68 training examples. We then ran a comparative study to assess both user experience (what it is like to use) and effectiveness (what people learn). The user experience was acceptable overall, with clear room for improvement. We could not draw firm conclusions about effectiveness, but our results suggest that newcomers can gain introductory knowledge of computational thinking.
[This summary has been rewritten with the help of AI based on the project's original abstract]
Keywords
Documents