Analyzing the Evolution of Digital Assessment in Education Literature Using Bibliometrics and Natural Language Processing

Introduction

Over the last two decades, technology has become an inherent part of education, by generating multiple new applications to improve the learning process (Raja & Nagasubramani, 2018) . These applications have been diverse, including learning management systems (LMSs) to facilitate course development (Sclater, 2008), smart devices and classrooms (Zhu et al., 2016), artificial intelligence applications in education (L. Chen et al., 2020), or even games (Ruipérez-Valiente & Kim, 2020). Multiple research studies have shown the benefits of educational technologies to improve learning. Within the broad spectrum of technology enhanced learning, researchers focus on this chapter specifically on digital assessment, also known as e-assessment (Whitelock, 2009).

Broadly speaking, assessment can be defined as a process of drawing inferences based on evidence. Within the context of education, it has been defined in different ways. Our view of assessment aligns with the following definition provided by Huba and Freed “Assessment is the process of gathering and discussing information from multiple and diverse sources in order to develop a deep understanding of what students know, understand, and can do with their knowledge as a result of their educational experiences” (Huba & Freed, 2000). Therefore, when we talk about digital assessment, we mean that this assessment process is supported by digital technologies at some point. In that sense, digital assessment can take many forms. The simplest form might be the use of digital quizzes and exams in order to facilitate performing assessment over a digital medium (Dellos, 2015). However, that is just the very first step in the process.

Over the last decade there have been multiple digital assessment approaches aiming to improve the educational process. For example, numerous intelligent tutoring systems (ITSs) (Anderson et al., 1985) have emerged that frequently implement adaptive learning algorithms in order to adapt the assessment items to the current knowledge of students automatically. Other examples can include the use of health assessment systems (Saini et al., 2012) to evaluate the recovery of patients that are training to get better or the implementation of digital assessment solutions that can improve the trustworthiness of remote learning against academic dishonesty (Jaramillo-Morillo et al., 2020). Finally, a prominent example is the field of game-based assessment, that aims to perform stealth assessment of competencies and skills through the use of the data generated in games (Gomez et al., 2021). These different applications have reported clear benefits that can improve the assessment process at different levels. Some studies are focused on reporting new tools for digital assessment, others focus on the instructional design to include those digital assessments within the curriculum, and others evaluate the outcomes of using digital assessment.

Therefore, the field of digital assessment is quite broad. In this chapter, we aim to perform a longitudinal study of this literature. However, performing a qualitative review of all the literature is quite consuming, given that this is an ample field with many publications. Consequently, we propose to perform a bibliometrics study enhanced with natural language processing (NLP) techniques (Wolfram, 2016) to automatically extract the topics of the papers based on their full text or abstract. We will investigate the main topics and the evolution of those over the last 15 years. Moreover, we also aim to use network analysis approaches to inspect the research community with a co-author network and a citation paper network. More specifically, we have the following research questions (RQs):

RQ1: What are the main topics of digital assessment in education literature based on keywords and topic modeling?

RQ2: What has been the evolution of such topics over the last 15 years of research?

RQ3: What are the communities of authors and papers in this topic based on a network analysis perspective?