Automatic Item Generation with Reading Passages: A Systematic Literature Review

Automatic Item Generation (AIG) is a system to generate items/questions automatically with a model using computer technology. AIG can help teachers to provide questions in education assessments more efficiently. Currently, AIG is already implemented in various educational subjects. Based on the author’s knowledge, there are no review papers that discuss AIG with reading passages. Here we conduct a systematic literature review to provide a comprehensive review of AIG, especially AIG with reading passages. The research objective is to know about AIG approaches datasets, evaluations, and challenges. We use S phases: planning and managing the review, defining the research question, identifying inclusion and exclusion criteria, search strategy, screening titles and abstracts, selecting full-text paper, data extraction, analysis, and synthesis. This research reviewed 30 relevant papers between 201S and 2022. We found that various approaches have been implemented. An approach is not driven by what kind of language will be used in the generated question. On the other hand, dataset selection is always based on the question language and question type that will be generated. Before generated questions are used in assessment, it is necessary to evaluate the question quality. Despite the success stories above, AIG still has some challenges for future development.