Aspect-based sentiment analysis of reviews in the domain of higher education
DOI | https://doi.org/10.1108/EL-06-2019-0140 |
Date | 27 January 2020 |
Pages | 44-64 |
Published date | 27 January 2020 |
Author | Nikola Nikolić,Olivera Grljević,Aleksandar Kovačević |
Subject Matter | Information & knowledge management,Information & communications technology,Internet |
Aspect-based sentiment analysis
of reviews in the domain of
higher education
Nikola Nikoli
c
Department of Applied Computer Science and Informatics,
University of Novi Sad Faculty of Technical Sciences, Novi Sad, Serbia
Olivera Grljevi
c
Department of Business Informatics and Quantitative Methods,
University of Novi Sad Faculty of Civil Engineering Subotica, Subotica, Serbia, and
Aleksandar Kova
cevi
c
Department of Applied Computer Science and Informatics,
University of Novi Sad Faculty of Technical Sciences, Novi Sad, Serbia
Abstract
Purpose –Student recruitment and retention are important issues for all higher education institutions.
Constant monitoring of student satisfaction levels is therefore crucial. Traditionally, students voice
their opinions through official surveys organized by the universities. In addition to that, nowadays,
social media and review websites such as “Rate my professors”are rich sources of opinions that should
not be ignored. Automated mining of students’opinions can be realized via aspect-based sentiment
analysis (ABSA). ABSA s is a sub-discipline of natural language processing (NLP) that focusses on the
identification of sentiments (negative, neutral, positive) and aspects (sentiment targets) in a sentence.
The purpose of this paper is to introduce a system for ABSA of free text reviews expressed in student
opinion surveys in the Serbian language. Sentiment analysis was carried out at the finest level of text
granularity –the level of sentence segment (phrase and clause).
Design/methodology/approach –The presented system relies on NLP techniques, machine
learning models, rules and dictionaries. The corpora collected and annotated for system development
and evaluation comprise students’reviews of teaching staff at the Faculty of Technical Sciences,
University of Novi Sad, Serbia, and a corpus of publicly available reviews from the Serbian equivalent
of the “Rate my professors”website.
Findings –The research results indicate that positive sentiment can successfully be identified with the
F-measure of 0.83, whilenegative sentiment can be detected with the F-measure of 0.94.While the F-measure
for the aspect’s range is between0.49 and 0.89, depending on their frequency in the corpus. Furthermore,the
authors have concluded that the quality of ABSA depends on the source of the reviews (official students’
surveys vs reviewwebsites).
Practical implications –The system for ABSA presented in this paper could improve the quality of
service provided by the Serbian higher education institutions through a more effective search and
summary of students’opinions. For example, 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
servicemoreattentionshouldbedirected.
Originality/value –To the best of the authors’knowledge, this is the first study of ABSA carried out at
the level of sentencesegment for the Serbian language. The methodology and findings presentedin this paper
Results presented in this paper are a part of the research conducted within the grants III-47003 and
III-44010 provided by the Ministry of Education and Science of the Republic of Serbia.
EL
38,1
44
Received8 June 2019
Revised28 October 2019
Accepted26 November 2019
TheElectronic Library
Vol.38 No. 1, 2020
pp. 44-64
© Emerald Publishing Limited
0264-0473
DOI 10.1108/EL-06-2019-0140
The current issue and full text archive of this journal is available on Emerald Insight at:
https://www.emerald.com/insight/0264-0473.htm
provide a much-needed bases for further work on sentiment analysis for the Serbian language that is well
under-resourcedand under-researched in this area.
Keywords Digital documents, Higher education, Document management, Machine learning,
Opinion mining, Document handling, Aspect-based sentiment mining, Automatic indexing,
Semantic document indexing, Student reviews
Paper type Research paper
1. Introduction
Educational institutions around the world spend a lot of money to improve conditions and
services, retain existing studentsand recruit new ones. According to a survey conducted in
2018 in the USA (Levitz, 2018), a four-year public university spends an average of $536,
while a private university spends four times more for recruiting a student. Therefore,
student opinions and satisfaction with the educationalinstitution’s service quality are very
important, and it represents a type of marketing, more commonly known as “word-of-
mouth”(WoM) (Li, 2013) and “electronicword-of-mouth”(eWoM) (Yahya et al., 2014), which
significantly affectsstudent recruitment and faculty reputation (Rauschnabelet al., 2016).
Students express their opinions through official surveys and questionnaires conducted at
the institutional level and online through social networks and forums. Most of the official
surveys and questionnaires provide an option of leaving a free text (unstructured) review.
Those reviews, along with online reviews, carry useful information and need to be analysed.
Typically, unstructured reviews are indexed and searched using key terms and standard
information retrieval (IR) techniques. However, a search by key terms is not an easy way to
retrieve reviews where students express a certain sentiment polarity (positive, negative,
neutral) towards a particular aspect of education (professor, course organization, lecture, etc.).
Indexing unstructured reviews by aspect and sentiment polarity would enable a m ore efficient
search and summary of student opinions. In this way, the educational institution’s
administration couldeasily identifywhich aspects oftheir educationalexperience the students
are not satisfied with and for which aspects they will spread negative WoM and eWoM.
To index reviews by aspect and sentiment, it is necessary to annotate them with
appropriate aspects and assign each of the aspect a sentiment polarity. Manual annotation of
sentiment and aspects is impractical because of the large number of documents. Therefore,
automatic processing is required. Automatic extraction of sentiments and aspects from
unstructured content is called “aspect-based sentiment analysis”(ABSA), and it is a sub-
discipline of sentiment analysis (opinion mining) (Liu, 2017;Thet et al., 2010). Despite a large
number of systems and methodologies for ABSA –for example, for social networks (Zhang
et al., 2018;Zhou and Zhang, 2017) and movie reviews (Thet et al., 2010;Trivedi et al., 2018)–
they are domain- and language-dependent (predominantly developed for the English language).
The majority of the ABSA systems operateat the sentence level. However, sentence-level
systems cannot handle cases where one sentence contains multiple aspects and sentiments
(e.g. “The professor is excellent, but the coursematerials are poorly organized.”). Therefore,
an efficient ABSA system for student reviews should be adapted to the domain of higher
education, which will be able to resolve multi-aspect sentences (and obviously consider the
language used in the reviews).
This paper introducesa sentence segment level (phrase or clause) ABSA systemadapted
to the domain of higher education in the Serbian language. The system comprises multiple
steps. Firstly, a review is splitinto sentences and then phrases andclauses using a custom
splitter built for the purposeof this research. Each of the segment is then tagged with oneof
the pre-defined aspects by a module consisting of multiple machine learning (ML) models.
Aspect-based
sentiment
analysis
45
To continue reading
Request your trial