A Review of Recent Advances on Automatic Question Generation for the Portuguese Language

Abstract

Although research of question generation has seen significant advancements, particularly in English and other high-resource languages, this trend is not perceived in the vast majority of low-resource languages, such as Portuguese. These low-resource languages do not benefit from abundant resources, such as large datasets, robust computational tools, and extensive linguistic studies. This, in turn, hinders and slows the progress of various fields in tasks that require support for those specific languages. In this survey, we aim to bridge this gap by providing a comprehensive summary and review of the field's current status, examining the methodologies and findings of recent studies. By reviewing 117 papers, we selected 10 studies that fit our inclusion criteria to identify gaps in the field and present recommendations for future research. We highlight the challenges and opportunities for developing AQG in Portuguese, a crucial area for advancing education in Portuguese-speaking countries. We propose that future research prioritize the development of more comprehensive datasets and explore ways to reduce computational costs.