Argumentative Zoning of Student-Written Thesis Abstracts in Latvian and English

This corpus consists of AZ-annotated bachelor theses abstracts, each written in Latvian and English. The abstracts are annotated using a modified version of the Swales’ CaRS model.The corpus was constructed by randomly sampling 1,252 pairs abstracts in English and Latvian from a total of 1,986 pairs extracted from bachelor’s theses defended at Riga Technical University during 2023–2024. The theses span a wide range of scientific disciplines, including economics, computer science, civil engineering, mechanical engineering, chemistry, telecommunications, architecture, and others. The abstracts were extracted directly from the theses and include supplementary elements such as keyword lists, thesis length information, and bibliographic data. It should be noted that all abstracts were already publicly available via the Riga Technical University website (https://nda.rtu.lv/en/). Software developed for dealing with this data: https://github.com/GintsJekabsons/AZ