Thematic Analysis of Indonesian Physics Education Research Literature Using Machine Learning

Abstract

Abundant physics education research (PER) literature has been disseminated through academic publications. Over the years, the growing body of literature challenges Indonesian PER scholars to understand how the research community has progressed and possible future work that should be encouraged. Nevertheless, the previous traditional method of thematic analysis possesses limitations when the amount of PER literature exponentially increases. In order to deal with this plethora of publications, one of the machine learning (ML) algorithms from natural language processing (NLP) studies was employed in this paper to automate a thematic analysis of Indonesian PER literature that still needs to be explored within the community. One of the well-known NLP algorithms, latent Dirichlet allocation (LDA), was used in this study to extract Indonesian PER topics and their evolution between 2014 and 2021. A total of 852 papers (~4 to 8 pages each) were collectively downloaded from five international conference proceedings organized, peer reviewed, and published by Indonesian PER researchers. Before their topics were modeled through the LDA algorithm, our data corpus was preprocessed through several common procedures of established NLP studies. The findings revealed that LDA had thematically quantified Indonesian PER topics and described their distinct development over a certain period. The identified topics from this study recommended that the Indonesian PER community establish robust development in eight distinct topics to the present. Here, we commenced with an initial interest focusing on research on physics laboratories and followed the research-based instruction in late 2015. For the past few years, the Indonesian PER scholars have mostly studied 21st century skills which have given way to a focus on developing relevant educational technologies and promoting the interdisciplinary aspects of physics education. We suggest an open room for Indonesian PER scholars to address the qualitative aspects of physics teaching and learning that is still scant within the literature.

An Online Physics Laboratory Delivered Through Live Broadcasting Media: A COVID-19 Teaching Experience

“…On the other hand, physics students still experienced learning obstacles in the context of graphical representation. In fact, this ability is imperative for physics conceptual understanding and problem solving (Docktor & Mestre, 2014;Hidayatulloh et al, 2021;Odden et al, 2020;Santoso et al, 2022). To address this issue, instructors should empower students to acquire this skill, using teaching tools such as scaffolding (Rangkuti & Karam, 2022), contextual tasks (Scheid et al, 2019), and obviously, as we discovered in our third meeting, tutorials (Kohnle & Passante, 2017).…”

“…Such disruption can influence the effectiveness of physics labs. Reporting on the effectiveness of physics instruction is broadly acknowledged as the main intention of physics education research (PER; Docktor & Mestre, 2014;Odden et al, 2020;Santoso et al, 2022). In this study, students' performance and their attitude to the disruption were proxies to evaluate the effectiveness of online nuclear physics laboratories during the pandemic.…”

Using Multilevel Modelling to Evaluate Science Literacy and Technology Course of the Indonesian Non-Science Students

“…Many research works have been attempted to develop scientific literacy worldwide by the science education research community (Impey et al, 2011;Queiruga-Dios et al, 2020;Odden et al, 2021;Santoso et al, 2022). Survey reports established by the Programme for International Student Assessment (PISA) and the Trends in International Mathematics and Science Study (TIMSS) undoubtedly impact educational initiatives worldwide, including the Indonesian scien-ce education system (Cheema, 2017;Rachman et al, 2021;Ustun et al, 2022).…”

Improving Spell Checker Performance for Bahasa Indonesia Using Text Preprocessing Techniques with Deep Learning Models

“…Stemming removes prefixes or suffixes from words to revert them to their base forms, while lemmatization transforms words into their basic forms so that variations of a word with the same meaning are standardized [19,20]. The quality of the final model is influenced by the preprocessing technique used [21,22].…”