Аннотация
The digitalization of secondary education in Kazakhstan increases the need for specialized Kazakhlanguage resources that support intelligent question-and-answer systems for school subjects. However, existing QA datasets are mainly focused on high-resource languages and do not cover the content of academic disciplines, particularly the history of Kazakhstan. This paper presents and analyzes in detail the Kazakh HistoryQA corpus, the first specialized question-and-answer resource in the Kazakh language, which is based on six official textbooks on the history of Kazakhstan for grades 5-10. The corpus includes 208 thematic sections, about 180 thousand tokens, and 661 QA-pairs with exact coordinates of the response span, which ensures compatibility with standard machine reading model training pipelines, including the HuggingFace library. A multi-level analysis of the corpus has been conducted: statistics on the length of contexts, questions, and answers, distribution of materials by classes and textbooks, and typology.