Making a difference in education with technology.
Individual students have different aptitudes and background that determine their learning needs. To improve the learning process, we want to develop a system that provides students with the flexibility to pick up new skills and knowledge in a more adaptive manner customised to their individual learning needs.
In this project, AICET collaborates with Ministry of Education (MOE) to develop and deploy a new adaptive learning system to support self-directed learning on the Singapore Student Learning Space (SLS).
Examination marking is a common task in education. Transporting scripts and ensuring that every script is accounted for involves a lot of manual work and educators are prone to mistakes. For examinations involving a large number of students, scripts are often marked by a team of examiners, each in charge of specific questions. Softmark allows scripts to be digitalised and made available for online marking, so educators no longer have to worry about losing scripts and can mark the same scripts in parallel, leading to greater efficiencies. The scripts can be marked anywhere and at any time with minimal restrictions.
Computer vision algorithms are integrated to tag scanned scripts to students and to automatically grade multiple choice questions. In the longer term, Softmark will be extended to support the typesetting of examinations so that the whole exam setting and marking process can be fully integrated end-to-end.
Personalised Timelines for Adaptive Learning
It is well known that students learn at different speeds. Classes are generally taught at the pace for all students and it is inevitable that some will fall behind. Classical adaptive learning approaches attempt to cater to different learning paces by providing students with content and questions that are best suited at their assessed level of mastery. This is extremely costly in terms of content creation and cannot be broadly applied without significant resources.
We have pioneered a new approach to adaptive learning where instead of providing different learning resources and content to different students, we provide the same set of well-curated and tested learning resources to all students. Students are differentiated instead by the pace of learning with their own personalized learning timelines. Faster students will be able to complete the materials to be covered faster and be able to move on to more advanced materials as soon as they are done, while slower students will be given more time to learn and complete their assignments.
Our new approach has been implemented and deployed on Coursemology, a gamified e-learning platform. Results seem promising but more work needs to be done to develop more advanced techniques in modelling student learning and in controlling the pace of learning. We also need to develop a visualisation dashboard to allow educators to understand the progress of a class of students and to be able to make adjustments to the class schedules for individual students on-the-fly in an intuitive and principled way. This is still work in progress, but we believe that once perfected, our approach will allow educators worldwide to provide affordable adaptive and personalized learning to their students.
Current systems for the automated grading of computer programming provide feedback on the correctness of the submitted programmes. They are, however, unable to provide feedback to the students on how they can improve their codes if they get the answers wrong without the intervention of human instructors. To improve learning efficiency, it is important for students to be able to get immediate feedback so that they are able to correct the wrong lines and arrive at the coding solution to deliver the desired outcomes.
In this project, we will develop techniques to provide automated feedback for programming questions using a novel combination of programming languages (PL) and machine learning (ML) techniques. The feedback will be able to identify the errors made by students during programming and provide a brief explanation on why the lines are wrong. The feedback can also give a hint on how the students can correctly modify the lines on their own without waiting for their lecturers or tutors to provide an answer, and this will lead to self-directed learning.
Choose Your Own Narrative (CYON):
Adaptive Self-Directed Learning System
Current teaching approaches mostly assume that students are the same and they are expected to learn at the same pace and have the same background. Educators typically use the same set of materials for all students in a course of study. Generally, students are guided by a single "narrative" throughout his or her learning journey. Such a fixed teaching system does not provide the better students with the flexibility to learn faster, and it might also cause distress to weaker students who are struggling to cope with the standard pace.
An adaptive learning approach can allow us to cater to individual student's needs.
In this project, we investigate a new approach to adaptive learning that we call Choose Your Own Narrative (CYON). CYON provides multiple pathways for the students to learn at his or her own pace. Our intelligent tutor then suggests potential pathways to reach the learning objectives by analysing the observed students’ preferred learning approaches and performance. In addition to supporting instructor-defined pathways, CYON also allows students to select their own learning materials, reference sites and even assessment tasks. These are in turn offered to other students as potential alternative pathways. Through monitoring and analysing of the student users’ selections, we expect well-trodden learning paths to emerge naturally from the wisdom of the crowd. This will redefine the way to learn a particular topic, enabling the students to potentially choose the more effective "narrative" for their own learning journey.