Question Assessment Recommendation System Based on Personalization using Collaborative Filtering and Long-Short Term Memory

The use of e-learning has the potential to change the way students learn, and it can be further improved by utilizing stored digital data to provide an engaging and interactive learning experience. The Collaborative Filtering approach is applied to increase student learning interest by providing personalized questions that are tailored to each student’s needs and abilities. This feature is used to solve teacher issues and the need to create assessments that be time-consuming and not accustomed to digitalizing questions. During creating an assessment, the system will provide the teacher with recommendation questions; therefore, teacher is able to adjust it based on every curriculum. Another application of machine learning is using Long Short-Term Memory helps the teacher by being able to forecast student scores and is used as an early detection tool for identifying students who may be struggling or losing interest in learning. The data source obtained from e-learning to implement both machine learning algorithms is limited to data on students’ assessment scores and data on questions with mathematics subjects. This work is intended to assist primary school mathematics teachers in making personalized assessments for each student with the aim of presenting interactive learning media as well as being a medium for monitoring student performance.