Jadwal Sholat

Memuat jadwal sholat…

Pendidikan editorial

Open AccessOA2020

On the Use of Soft Computing Methods in Educational Data Mining and Learning Analytics Research: a Review of Years 2010–2018

This review surveys 300 journal articles (2010–2018) that apply soft computing methods to educational data mining and learning analytics. It finds that quality-related issues dominate research goals, and identifies decision trees, neural networks, fuzzy logic, SVMs, and evolutionary algorithms as key techniques.
A. Charitopoulos; M. Rangoussi; D. Koulouriotis· International Journal of Artificial Intelligence in Education· 2020· DOI 10.1007/s40593-020-00200-8

The core problem

Educational Data Mining (EDM) and Learning Analytics (LA) are closely related research areas that analyze data generated by interactive/e-learning systems to address open education-related questions. Such systems store large volumes of data that can be exploited to assess the learner, the system, and the quality of interaction. Besides classic data analysis methods (clustering, classification, identification, regression/analysis of variances), soft computing methods are often employed. Their iterative optimization nature—avoiding exhaustive search and accepting possibly suboptimal solutions at realistic time and effort—along with their reliance on rich data sets for training, makes them ideal for EDM/LA problems. Decision trees, random forests, artificial neural networks, fuzzy logic, support vector machines, and genetic/evolutionary algorithms are examples that handle uncertainty, qualitatively stated problems, and incomplete, imprecise, or contradictory data sets—features common to education and social sciences. This review focuses on recent EDM and LA research that employs at least one soft computing method, aiming to identify: (i) major education problems/issues addressed and research goals/objectives set, (ii) learning contexts/settings, (iii) relation between classic and soft computing methods, and (iv) means of dissemination (publication journals).

Innovation

The authors conducted a systematic review of 300 journal publications from 2010–2018 that use soft computing methods in EDM and LA research. The selection process involved identifying relevant articles from journals and analyzing them according to the four research objectives. The review categorized publications by educational problem addressed, learning context, methods used (both classic and soft computing), and publication venue. The analysis focused on extracting patterns and trends across the body of literature, including frequency of methods and tools, and the relationship between problem types and methodological choices. The authors also examined the dissemination of research results through journals to understand where such research is published worldwide.
Introduction
Educational Data Mining (EDM) and Learning Analytics (LA) are closely related research areas that analyze data generated by interactive/e-learning systems to address open education-related questions. Such systems store large volumes of data that can be exploited to assess the learner, the system, and the quality of interaction. Besides classic data analysis methods (clustering, classification, identification, regression/analysis of variances), soft computing methods are often employed. Their iterative optimization nature—avoiding exhaustive search and accepting possibly suboptimal solutions at realistic time and effort—along with their reliance on rich data sets for training, makes them ideal for EDM/LA problems. Decision trees, random forests, artificial neural networks, fuzzy logic, support vector machines, and genetic/evolutionary algorithms are examples that handle uncertainty, qualitatively stated problems, and incomplete, imprecise, or contradictory data sets—features common to education and social sciences. This review focuses on recent EDM and LA research that employs at least one soft computing method, aiming to identify: (i) major education problems/issues addressed and research goals/objectives set, (ii) learning contexts/settings, (iii) relation between classic and soft computing methods, and (iv) means of dissemination (publication journals).

Why it matters

The review discusses weaknesses and issues that need further attention to fully exploit the benefits of research results to improve both the learning experience and learning outcomes. It notes that while soft computing methods are well-suited to educational problems due to their ability to handle uncertainty and imprecise data, there are challenges in their application. The authors emphasize the need for continued methodological development and better integration of research findings into practice. The relation between classic and soft computing methods is examined, showing that soft computing approaches are often used in combination with classic methods to address complex educational issues. The discussion also points to the dominance of quality as a research focus, reflecting the shift towards e-learning and technology-mediated learning environments.

Who should read this

Pembaca Research Digest dan praktisi rumpun ini.

Opening member content…