Abstract
In the context of smart classrooms (SC), one of the sources that can enrich data collection, analysis, real-time feedback, and decision making are students' emotions. This research tried to analyse the knowledge published over the last 5 years about the effectiveness of emotion recognition (ER) interventions in classrooms. A total of 214 articles were chosen based on the search terms and analysed according to the PRISMA statement, and finally 39 were selected. The findings of the interpretation of facial image-based ER have been upgraded with rapid and power progress of deep learning technology. As emotions can be detected using different sort of input, such as speech, facial expressions, videos, messages, and emoticons, the main point is tracking emotions while the lesson is taking place so as to warn the teacher. It is of utmost interest if one seeks to improve the student's academic performance, improve teaching, and understand students' learning behaviour.
TopBackground
Asking ourselves about the main thread of the research: Which are the emotions? How can emotions be detected? How can they affect teaching and learning? How is ER applied in the classroom? In order to reply these research questions, we focus on the next research objectives: Identifying the emotions and the main studies around ER based on AI in the last five years (2017-2021), using a systematic review; also analysing any educational assessment which is being improved with ER.
Therefore, this research is based on the use of a systematic review as a method to answer a research question through a systematic and replicable process. Using for this review, an announcement model called PRISMA.
The present research identified 39 articles that dealt with this. Such models with the collaboration of technology allow us to advance and detect what is described.
The literature search identified from consideration of over 214 papers, with only 39 articles based on the search terms, which includes information on 23 references of relevance.
The vast majority of these papers were excluded due to one of the following reasons: were not relevant in an educational context; were focused on ER but only for seeing how students with special needs feel; were related to facial recognition so as to check attendance in class, also were linked with ER but exclusively on e-learning, not face-to-face in a classroom. Eventually, 39 articles were selected for review and uploaded to Mendeley for further research. In the subsequent sections, the articles have been examined according to the emotions, the psychology and the interpretation of these emotions, the learning status of students, and the effectiveness of interventions based on ER, all of them with the same common goal, to motivate, to engage the students and to involve them better into the learning process, improving their academic performance, as has been said, through studying students' emotions.
This study was first collected via two different reputable citation databases: Scopus and Web of Science, using the key terms for filtered 5 past years, that is, from 2017 to 2021. Most of the articles were published in 2021 (3), followed by 9 articles in 2020, 11 in 2019, 6 in 2018, and 3 in 2017. The remaining articles (7) are those derived from the search in the article references, and are from previous years.
Key Terms in this Chapter
Artificial Intelligence algorithms: Indicates to the computer how to operate on its own as it learns and improves.
E-Learning: A way of getting knowledge through information and communication technologies.
Facial Recognition: A technology capable to match from a database of a human face to a digital or video image.
Portable wireless devices: Battery-powered autonomous with no wire devices.
Melspectrogram: Represents a sound, an acoustic time-frequency.
Wearables: A kind of electronic devices worn as accessories that are embedded in clothing.