Meta-Heuristic Algorithms for Learning Path Recommender at MOOC

Abstract

Online learning platforms, such as Coursera, Edx, Udemy, etc., offer thousands of courses with different content. These courses are often of discrete content. It leads the learner not to find a learning path in a vast volume of courses and contents, especially when they have no experience in advance. Streamlining the order of courses to create a well-defined learning path can help e-learners achieve their learning goals effectively and systematically. The learners usually ask the necessary skills that they expect to earn (query). The need is to develop a recommender system that can search for suitable learning paths. This study proposes a multi-objective optimization model as a knowledge-based recommender. Our model can generate an appropriate learning path for learners based on their background and job goals. The recommended studying path satisfies several learner criteria, such as the critical learning path, number of enrollments, learning duration, popularity, rating of previous learners, and cost. We have developed Metaheuristic algorithms includes the Genetic Algorithm (GA) and Ant Colony Optimization Algorithm (ACO), to solve the proposed model. Finally, we tested proposed methods with a dataset consisting of Coursera's courses and Vietnamwork's jobs. The test results show the effectiveness of the proposed method.

Metaheuristic Algorithms Based on Compromise Programming for Multi-Objective Urban Shipment Problem

“…We only use a single core to executes the algorithms. The executions can be speeded up using parallel mechanism for search agents proposed by Ngo et al [44]. Both GA and HGA algorithms use a similar search mechanism.…”

“…For this purpose, the weighted 𝐿 𝑝 metrics measure the distance of any solution from the reference point. The ideal objective vector is often used as the reference point: There are many studies that have used CP to approach the MOP problem such as for university timetabling [40], [41], [42], or in team selection [43], in knowledge based recommender [44], project task assignment [45]. However, it may require pre-defined minimal and maximal values of the objective functions.…”

“…Some of these values are predictable [43], most of other cases require to solve the problem as single objective function multiple times, which may be costly. Other studies [44] [46] show that the referential point may be selected from business estimations could provide better performance for the agents in searching process. However, in this study, the normalization method using both of 𝑧 * and 𝑧 𝑤𝑜𝑟𝑠𝑡 brings a better quality of the solution.…”

Genetic Algorithm for Solving Multi-Objective Optimization in Examination Timetabling Problem

“…In the future, we continue to improve the algorithm to increase population diversity [24] and improve scaling for the target functions to balance their importance. Improving computing performance with parallel computing is also one of our priorities [25], [26]. 6…”

Metaheuristic Algorithms Based on Compromise Programming for the Multi-Objective Urban Shipment Problem

“…For this purpose, the weighted 𝐿 𝑝 metrics measure the distance of any solution from the reference point. The ideal objective vector is often used as the reference point: Many studies have used CP to approach the MOP problem, such as for university timetabling [40][41][42], team selection [43], in a knowledge-based recommender [44], and project task assignment [45]. However, this t may require pre-defined minimal and maximal values of the objective functions.…”