Original Research

Automatic assessment of online discussions using text mining

Yvette Awuor, Robert Oboko
International Journal of Machine Learning and Applications | Vol 1, No 1 | a2 | DOI: https://doi.org/10.4102/ijmla.v1i1.2 | © 2012 Yvette Awuor, Robert Oboko | This work is licensed under CC Attribution 4.0
Submitted: 23 March 2012 | Published: 29 May 2012

About the author(s)

Yvette Awuor, Kenya Methodist University,
Robert Oboko, University of Nairobi, Kenya


Share this article

Bookmark and Share

Abstract

Online discussion forums have rapidly gained usage in e-learning systems. This has placed a heavy burden on course instructors in terms of moderating student discussions. Previous methods of assessing student participation in online discussions followed strictly quantitative approaches that did not necessarily capture the students’ effort. Along with this growth in usage there is a need for accelerated knowledge extraction tools for analysing and presenting online messages in a useful and meaningful manner. This article discussed a qualitative approach which involves content analysis of the discussions and generation of clustered keywords which can be used to identify topics of discussion. The authors applied a new k-means++ clustering algorithm with latent semantic analysis to assess the topics expressed by students in online discussion forums. The proposed algorithm was then compared with the standard k-means++ algorithm. Using the Moodle course management forum to validate the proposed algorithm, the authors show that the k-mean++ clustering algorithm with latent semantic analysis performs better than a stand-alone k-means++.

Keywords

k-means++ clustering; latent semantic analysis; singular value decomposition; text mining

Metrics

Total abstract views: 7937
Total article views: 17101

Reader Comments

Before posting a comment, read our privacy policy.

Post a comment (login required)

Crossref Citations

No related citations found.