Abstract:
Recommendation systems are now commonly used in e-learning to suggest
resources and learning materials to learners and improve teaching and learning quality.
To this end, predicting learners' needs and recommending e-learning resources in e-
learning systems has become a research focus. One way to address the need for
predicting user needs and improving the usability of e-learning systems is to
recommend pages or resources to learners that are related to their interests at a given
time. The purpose of this study is to evaluate the effectiveness of bundle correctness
percentages and frequencies in a web-based e-learning system for analysing learner
interests and recommending e-learning resources. The primary data source for this
research is EdNet. The primary distinguishing factor of this study is the organization of
some of the dataset's questions into bundles (groups of questions that must be answered
collectively). This dataset has about 95,293,926 interactions, 13,169 questions, 784,309
students, and 188 knowledge components. The approach involves clustering learner
sessions based on the accuracy of learning relevant bundles in a pattern specified by
our content-based recommender tool. The outcome of this study is a graphical user
interface. Users will provide the bundle they want to study, and our system will
recommend the prerequisite bundles required to understand the intended bundle. Out of
a total of 9,534 bundles, 8,935 are considered suitable learning objects, as 60% of
students can respond to the questions correctly. An efficient pattern for studying these
bundles is recommended as a study plan by a content-based education recommender
tool, which can further enhance user performance.
