Abstract:

E-Portfolio has recently become important in education.E-Portfolio systems gather amount of data such as learners output, performance, grade, learning diary and history.This huge amount of data has high potential to become a useful educational tool.

However, it is difficult to find beneficial e-Portfolio for learners from the massiveness of learners' data. For this reason, this paper proposes an e-Portfolio recommendation system that recommends learning output of learners who have similar interests in their e-Portfolios.This research consists of a topic model, Latent Dirichlet Allocation(LDA), a document classification method to analyze learning output in e-Portfolios. LDA can estimate topics, similar to categories or themes mainly included in documents. The proposed system computes similarity between learning output in e-Portfolios using topics.That recommends e-Portfolio for learners, which has similar topics.This paper demonstrates the effectiveness of the proposed system using actual classroom data in engineering course.

Keywords:

E-Portfolio, Recommendation System, Topic Model, Latent Dirichlet Allocation.