Sentiment Analysis in Education: ABSA and Student Feedback

Introduction

Teacher effectiveness can be gauged by analysis of student sentiment as communicated in written texts, particularly in social media posts (Misuraca, Forciniti, Scepi & Spano, 2020). Hajirizi and Nuci (2020) state that "being able to understand and find out what students like and don't like most about a course, professor, or teaching methodology can be of great importance for the respective institutions" (p. 1). For Kwecko et al. (2020), "the challenge is to understand the points of convergence and divergence between a set of opinions published on digital networks and their ability to reveal collective intelligences for the management of public policies in Education" (p. 2).

To overcome that challenge, both Hajirizi and Nuci (2020) and Nikolic, Grljevic, and Kovacevic (2020) recommend aspect-based sentiment analysis (ABSA) to help discern student sentiment. The use of ABSA appears to be a major trend in gauging sentiment analysis in education.

Sentiment Analysis

Misuraca et al. (2020) explain that processing data to engage in sentiment analysis requires "significant computational effort" (p. 1). As a sub-discipline of natural language processing, ABSA allows analysts to focus their attention on sentiments and their targets as revealed in a student's written or posted sentence. Yet as Nikolic et al. (2020) explain, students in higher education often communicate diverse sentiments with multiple targets in one sentence, and ABSA systems are generally not developed to withstand this level of complexity. For instance, if a student were to post — "The professor is excellent, but the course materials are poorly organized" — it would not be analyzed properly (Nikolic et al., 2020, p. 2).

To overcome this defect, Nikolic et al. (2020) developed an ABSA system that could break sentences into clauses and phrases by way of a custom splitter, so as to assign each clause or phrase a tag of positive or negative sentiment. This is just one example of how researchers are using technology to help benefit the overhaul of educational approaches. As Misuraca et al. (2020) put it, "determining students' views by collecting and processing feedback on their learning experiences is widely recognized as a central strategy for assessing the quality of teaching at most educational institutions" (p. 2). Through the text mining process — involving data collection and assembly, data processing, data exploration and visualization, model building, and model evaluation — an iterative approach to understanding student sentiment can be used to achieve this aim (Hajirizi & Nuci, 2020).

One obstacle that presents itself to this process, however, is the problem of subjectivity in texts: "the amount of subjective content existing in textual documents on the Web makes the activity of interpreting this type of information complex, aggravated when it is necessary to extract and classify subjectivity in a large volume of data from different sources" (Kwecko et al., 2020, p. 8). Thus, in spite of advances made in sentiment analysis, particularly with the model proposed by Nikolic et al. (2020), there are still obstacles to analysis that need to be overcome.

Sentiment analysis impacts learners and educators by helping educators understand students' views of courses and course material (Misuraca et al., 2020). With this information, educators can make adjustments to satisfy student needs as well as to address potentially negative student sentiment. If a course or its materials are found to be viewed negatively by students, educators can review their approaches or the way course material is presented and determine what can be done to improve the student experience. Feedback is important for educators because it helps them understand how students perceive their instruction.

How Sentiment Analysis Impacts Learners and Educators

Students can also benefit from this process, because by voicing their sentiments they can alter or reinforce the way a course is taught (Misuraca et al., 2020). As Nikolic et al. (2020) put it, "a particular educational institution could very easily find out which aspects of their service the students are not satisfied with and to which aspects of their service more attention should be directed" (p. 1). By bringing teacher views in line with student views, a more aligned perspective on education can be achieved.

Since the goal of education is to facilitate the learning process of the student, student sentiment should be taken into consideration. Sentiment analysis helps educators understand student sentiment in a way that was previously impossible. Thanks to students' use of social media to provide feedback and insight into their feelings, thoughts, views, and experiences, educators have access to data that allows them to see their approaches from the students' perspective. Whereas educators previously had to rely on student surveys to understand what students thought, they can now use sentiment analysis to obtain even deeper insight (Kwecko et al., 2020).

Conclusion

Regardless, sentiment analysis is a major trend that is developing in education, as educators seek to understand what their students are experiencing and willing to share outside of surveys, which can limit educators' perspective on student sentiment (Misuraca et al., 2020). The more that educators are able to apply advances in sentiment analysis — using techniques such as ABSA — the more likely they are to be able to augment their own approaches to education to help both themselves and their students achieve their goals.

Sentiment analysis helps researchers understand what people are thinking and feeling by analyzing texts published online, most notably via social media. In education, sentiment analysis is playing a key role in helping educators understand student sentiment regarding courses and course material. Until recently, educators have relied mainly on surveys to obtain student feedback. However, surveys limit insight because they do not generally give students an opportunity to voice matters in their own words.

Social media platforms do provide that opportunity, and sentiment analysis can be used to read student-generated text and determine whether students have a positive, negative, or neutral outlook on a particular course or approach to education. Educators can use this information to adjust or reinforce their own pedagogy, and students benefit because their voices are being heard and used to augment their education. The educators are there for the students, and sentiment analysis is a tool that is there for educators.

Hajrizi, R., & Nuçi, K. P. (2020). Aspect-based sentiment analysis in education domain. arXiv preprint arXiv:2010.01429.

Kwecko, V., de Tôledo, F. P., Devincenzi, S., Ortiz, J. O. D. S., & Botelho, S. S. D. C. (2020, October). Analysis of the feelings of the population's opinion in social media: A look at education. In 2020 IEEE Frontiers in Education Conference (FIE) (pp. 1–9). IEEE.

Misuraca, M., Forciniti, A., Scepi, G., & Spano, M. (2020). Sentiment analysis for education with R: Packages, methods and practical applications. arXiv preprint arXiv:2005.12840.

Nikolić, N., Grljević, O., & Kovačević, A. (2020). Aspect-based sentiment analysis of reviews in the domain of higher education. The Electronic Library.