PewePro

Adaptive Question Selector

A method for adaptive question selection based on student learning model.

Overview
Technologies used: Microsoft .NET, LINQ, Microsoft SQL Server
Inputs: Course data created in QTI Editor
Output: Recommended question or concept

Addressed Problems

Web-based learning and web learning systems are today very popular because they enable students to access course materials at any time from any place. Adaptive learning systems are capable to monitor and analyze user activity and adapt to his personal needs and interests.

Proposed method is based on evaluation of user knowledge using test questions which are inserted into learning materials. Using adaptive testing methods most appropriate questions for specific user are selected. It can also recommend appropriate topics for user which is also based on evaluation of his knowledge.

Description

Method can adapt to the user by selecting appropriate questions and recommending concepts to further study. Adaptive question selection is a three step method which uses data from user model (see figure below). In first step it selects most appropriate concept using prerequisite graph. Second step is based on Item Response Theory and selects defined number of questions with adequate difficulty. Third step is prioritizing questions using user history which contains time and correctness of each user answer.

method overview

Selection method overview.

After the question is selected it is presented to the user and correctness of his answers is evaluated. Then the user model is updated in each step of the method that is now executed in inverse order. Table below shows selection and update process in greater detail.

Order Step name Question selection Answer evaluation
1 Concept selection using course structure Selects the most appropriate concept of question Using user knowledge estimation as a result of second method it updates appropriate concept for user
2 Item Response Theory Selects k-best questions with most appropriate difficulty Updates the estimation of user knowledge of question concept
3 History based prioritization of questions Selects one question that was not recently asked or using other history based strategy Updates the log of time and correctness of user answer

Concept recommendation is based only on the first step of selection method which returns most appropriate concept to use.

Publications

  1. Vozár, O., Bieliková, M. Adaptive Test Question Selection for Web-based Educational System. In: Proc. of SMAP 2008 - 3rd Int. Workshop on Semantic Media Adaptation and Personalization. Prague, CR, 2008. CS IEEE Press, pp. 164-169.