This paper is a comprehensive examination of major learning styles and current approaches to automatically detect a student’s learning style. The introduction reports that over the last 20 years, researchers have identified 71 learning style models; however, there is a degree of overlap between these models. The paper evaluates five models that are commonly used for the automatic detection of learning styles: Kolb, Gardner, Felder, Biggs, and custom. The custom model is a collection of approaches that incorporates aspects from several traditional learning styles. The advantage of the custom model is being able to use multiple characteristics and learning style dimensions.

The two main approaches identified for the detection of learning styles are data driven and literature based. This classification is based on the types of techniques used to determine the learning style. The data-driven approach uses an artificial intelligence (AI) classification algorithm, while the literature-based approach applies a simple rule-based method from a user-defined model. The authors point out that their work is not meant to compare the approaches, but to highlight the performance and feasibility of each approach.

The discussion highlights the open issues that face this area of research and implementation. There are several topics of concern: the lack of cohesion and competing ideas between learning styles, research that utilized small data sets, and how differing research results can be properly evaluated. This paper has relevance to professional educators seeking insight into the topic, and would be of importance to online developers who are creating course content.