Architecture of a module for analyzing electronic test results

Abstract

This article presents the use of intelligent agents in analyzing the results of electronic tests, based on IMS Question & Test Interoperability (QTI) standard. The results of the analysis show the average score of the conducted test, sections of the study material and questions of the test that hinders students. The aim is to support the educational process by creating a personal assistant which is of service to teachers. The personal assistant is a part of a virtual learning space which consists of autonomous components with reactive, interactive and proactive behavior and are capable of maintaining different mental levels. The total mark of the test cannot be used to make substantiated conclusions regarding the understanding of the material. It is necessary to further analyze the data from each individual test from the conducted examination and also to analyze the student's activity during the process of training and their interim results. For this purpose a major role in the architecture of assistant takes a module whose main responsibility is to analyze each test separately and look for the cause of unsatisfactory results. This module is being developed as a multi-agent system of BDI agents. For presenting the internal states of the agents and formalization of beliefs, desires and intentions are used methods of commonsense reasoning and event calculus-based approach.