A low complexity heuristic to solve a learning objects recommendation problem

The recommendation of learning objects in virtual learning environments has become the focus of research to improve online learning experience. Several approaches have been presented in an attempt to model the individual characteristics of the students and offer learning objects that best suit their...

Full description

Saved in:
Bibliographic Details
Published in:Smart learning environments 2020-09, Vol.7 (1), p.1-17, Article 23
Main Authors: Falci, Samuel Henrique, Dorça, Fabiano Azevedo, Andrade, Alessandro Vivas, Mourão Falci, Daniel Henrique
Format: Article
Language:English
Subjects:
Algorithms
CAI
Complexity
Computational efficiency
Computer assisted instruction
Computers and Education
Computing costs
Distance learning
Education
Electronic Learning
Genetic algorithms
Heuristics
How to Make Learning Smart
Individual Characteristics
Integer programming
Learning object recommendation
Linear programming
Machine learning
Mathematical Applications
Mixed integer
Object recognition
Optimization
Social networks
Student Characteristics
Virtual environments
Virtual learning environment optimization
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The recommendation of learning objects in virtual learning environments has become the focus of research to improve online learning experience. Several approaches have been presented in an attempt to model the individual characteristics of the students and offer learning objects that best suit their particularities. Most of them, though, are impractical in real-world scenarios due to the high computational cost as a huge number of repositories offering learning objects such as Youtube, Wikipedia, Stackoverflow, Github, discussion forums, social networks and many others are available and each has a large amount of learning objects that can be retrieved. In this work, we propose a low complexity heuristic to solve this problem, comparing it to a classical mixed-integer linear programming model and classical genetic algorithm in varying dataset sizes that contain from 2000 to 1360000 learning objects. Performance and optimality were analyzed. The results showed that the proposed technique was only slightly suboptimal, while its computational cost was considerably smaller than the one presented by the linear optimization approach.
ISSN:2196-7091
2196-7091
DOI:10.1186/s40561-020-00133-8