Susman, M., Leijen, D. A. J., Johansson, C., Groom, N. (2022). Investigating Academic Document Structure using Object Detection Methods. Lecture Notes in Computer Science.
Abstract. The rhetorical structure of academic research papers written in English is now well understood, much less is known about the generic conventions governing academic texts written and published in less-studied languages. This article investigates the automatic detection of rhetorical patterns in academic texts using machine learning algorithms which were originally designed for image object detection purposes, and are thus entirely language independent. Our initial results indicate that this graphical, image-based approach to genre analysis is feasible. We intend to extend our approach to the detection of local variants and the rules of those variants.
