Reward based Erudition Platform Powered by Artificial Intelligence
¹ ABISHEK.C
¹ Student, Department of
Computer Science and Engineering
Kings College of Engineering,
Punalkulam, Pudukottai
abishekboss24@gmail.com
² NIRANJAN.R
² Student, Department of
Computer Science and Engineering
Kings College of Engineering,
Punalkulam, Pudukottai
niranjanvishwa74@gmail.com
³ NITHISKUMAR.V
³ Student, Department of
Computer Science and Engineering
Kings College of Engineering,
Punalkulam, Pudukottai
nithisn21@gmail.com
⁴ SATHISH KUMAR.S
⁴ Student, Department of
Computer Science and Engineering
Kings College of Engineering,
Punalkulam, Pudukottai
kumarsathish13475@gmail.com
⁵ M.ARUN
⁵ Assistant Professor, Department of
Computer Science and Engineering
Kings College of Engineering,
Punalkulam, Pudukottai
arun.cse@kingsengg.edu.in
Abstract
Online education has exploded in recent years, yet one frustrating problem has barely changed: learners still spend more time searching for the right course than actually learning. Platforms overwhelm users with thousands of options but offer little genuine guidance on where to start. Most recommendation systems either push whatever is trending or trust learners to accurately rate their own skill level — and neither approach works well. The result is predictable: wrong-level courses, early dropouts, and learning gains that fall well short of what they could be. This paper presents the Reward-Based Erudition Platform (RBEP), a system we built to tackle this problem head-on. The idea behind RBEP is straightforward: before recommending anything, actually measure what the learner knows. RBEP gives each learner a short diagnostic quiz and computes a reward score from three signals — how accurately they answered, how quickly they responded, and how consistent their performance was across difficulty levels. A hybrid machine learning engine then uses that score, along with a brief background profile, to recommend a course matched to the learner's actual ability. After the learner completes that course, a follow-up assessment checks whether the recommendation really paid off — and if it did not, the system tries again with a revised suggestion. Across our test set, RBEP reached 91.3% recommendation accuracy and a 78.4% completion rate, comfortably outperforming every baseline we tested.
KEYWORDS : Adaptive Learning, Artificial Intelligence, Course Recommendation, Reward-Based Learning, Machine Learning, Personalized Education, Learning Analytics, Decision Tree, Random Forest, K-Nearest Neighbors.
