토픽 모델링을 활용한 「교육평가연구」 학술지의 연구 동향 분석

Abstract

This study aims to explore research trends in educational measurement and evaluation by applying text-mining and analyzing the abstract of 969 articles published in the Journal of Educational Evaluation from 1993 to 2022, published by the Korean Society for Educational Evaluation. The main findings from keyword frequency analysis and topic modeling analysis based on LDA algorithm are as follows. First, keywords such as ‘student,’ ‘question,’ ‘test,’ ‘model,’ ‘evaluation,’ ‘estimation,’ ‘academic achievement,’ and ‘ability’ were generally mentioned as necessary in abstracts from the journal, and the main keywords changed over time. Second, a total of 12 topics appeared, including ‘the Item Response Theory model’, ‘development and validation of measurement tools’, ‘analysis of the factors influencing academic achievement’, ‘program evaluation and performance assessment’ and so on. Third, the ratio of the 12 topics that represent research trends changed with time. Especially, the topics of ‘analysis of factors influencing academic achievement’, ‘the Item Response Theory model’, and ‘teacher evaluation and student assessment’ had an increasing trend. In contrast, the topics of ‘computer-based evaluation’, ‘college entrance exam’, and ‘program evaluation and performance assessment’ had a decreasing trend. Lastly, the traits of research trends in the South Korean educational measurement and evaluation shown in the Journal of Education Evaluation were analyzed and future development and improvement tasks were discussed.