Nástroj pro automatické hodnocení písemných testů pomocí AI

Abstract

This thesis describes the architectural solution and implementation of an application for automatic test evaluation using artificial intelligence. The goal of this application is to simplify the process from creating a test to its evaluation, where teachers can easily prepare new tests and students can receive results and feedback immediately after completing the test. All open-ended questions will be assessed by language models consistently and independently. Teachers will be able to review and revise these evaluations or fully trust the artificial intelligence. Furthermore, this work presents an architecture based on microservices and describes the implementation, including all necessary diagrams. This application is well-suited for schools and universities, as it enables Lightweight Directory Access Protocol (LDAP) authorization, real-time communication, and other functionality essential for teaching.