Design and implementation of an AI chatbot for students
Ren, Xingyu; Xu, Hao (2026)
2026
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-202605049211
https://urn.fi/URN:NBN:fi:amk-202605049211
Tiivistelmä
The objective of this thesis project was to design and implement an artificial intelligence chatbot to provide immediate academic support for students in technical subjects. Cloud-based tools have data privacy risks and require an internet connection. The objective of this thesis project was to develop a safe, offline tutor to protect user data.
The system was built using the Python programming language and the Streamlit framework, applied a constructive research method. Qwen2.5-0.5B Large Language Model was selected for local calculations. The system was designed to process technical logic in English and then translate the output into Chinese for bilingual students. The testing strategy included automated interface checks and an evaluation with 20 technical questions.
The findings showed that the local model provided accurate academic support. It had an average response time of 4.10 seconds on an Apple M4 processor. The evaluation showed a maximum memory usage of 2.12 gigabytes. Our evaluation showed that the chatbot achieved an accuracy score of 4.8 out of 5.0. The study concluded that small open-source models can provide reliable educational support on standard computers without cloud services.
The system was built using the Python programming language and the Streamlit framework, applied a constructive research method. Qwen2.5-0.5B Large Language Model was selected for local calculations. The system was designed to process technical logic in English and then translate the output into Chinese for bilingual students. The testing strategy included automated interface checks and an evaluation with 20 technical questions.
The findings showed that the local model provided accurate academic support. It had an average response time of 4.10 seconds on an Apple M4 processor. The evaluation showed a maximum memory usage of 2.12 gigabytes. Our evaluation showed that the chatbot achieved an accuracy score of 4.8 out of 5.0. The study concluded that small open-source models can provide reliable educational support on standard computers without cloud services.