1 The United Graduate School of Education, Tokyo Gakugei University (JAPAN)
2 Center of Information and Communication Technology, Tokyo Gakugei University (JAPAN)
About this paper:
Appears in: EDULEARN19 Proceedings
Publication year: 2019
Pages: 9040-9049
ISBN: 978-84-09-12031-4
ISSN: 2340-1117
doi: 10.21125/edulearn.2019.2233
Conference name: 11th International Conference on Education and New Learning Technologies
Dates: 1-3 July, 2019
Location: Palma, Spain
It is becoming increasingly important not only in higher educational institutions but also in primary and secondary schools for students to learn proactively and autonomously. In this learning, it is essential for students to constantly reflect on their learning in order to adjust tasks and proceed to next learning.

There are as many different ways of reflecting on learning as there are students. For example, some students might stick to the facts and their impressions of various activities, while others might reflect on learning more deeply about the tasks they perform and the future prospects these open up. Thus, it is necessary for teachers to grasp the individual situation of each student, ideally through reading sentences of reflection on learning written by the student, and to facilitate reflection on learning appropriately.

However, it is not easy to grasp the individual situation of reflection on learning by reading the sentences. Therefore, it is thought that support for grasping the individual situation of reflection on learning is required. One promising direction is technologies such as artificial intelligence and machine learning, which could be used to facilitate learning by analyzing large amounts of accumulated data.

The purpose of this study is to develop a system that facilitates students reflect on learning by using a custom machine learning model. Our custom machine learning model aims to automatically classify written sentences of reflection on learning, with a focus on the different phases of reflection, and then grasps the situation of the individual student by using the classification result.

In our approach, we first need to grasp the individual situation of reflection on learning from their written sentences of reflection (requirement 1). Second, teachers need to facilitate further reflection on learning on the basis of the situation that has been grasped (requirement 2).

To satisfy the requirements, we designed following three functions. Function 1 automatically classifies written sentences of reflection on learning by means of a custom machine learning model in order to determine what kind of situation about reflection on learning the registered sentences belong in. This model is trained to classify sentences in accordance with specific phases of reflection on learning using data sets created in advance. Function 2 visualizes each student’s situation along with the registered sentences by using the results of function 1. Function 3 provides feedback to promote appropriate reflection on learning for students in accordance with the results of function 1.

In this study, we developed a student reflection support system with functions 1, 2, and 3 as a Web application. Two used cases are supported: one used by students and teachers and one used by students only. When used by students and teachers, teachers grasp each student’s situation and facilitate appropriate reflection by checking the results of classification (functions 1 and 2). When used by students only, the system (not the teachers) grasps each student’s situation and facilitates appropriate reflection by providing feedbacks (functions 1 and 3). Thus, by using our system, students will be able to reflect on learning appropriately and continuously regardless having to physically be in the classroom.

Evaluation results showed that our system could be used effectively in both use cases. In the near future, we will perform a more detailed evaluation.
Reflection on learning, learning record data, learning analytics, text classification, machine learning, e-learning systems.