KAZCAUSAL: ҚАЗАҚ ТІЛІНДЕГІ СЕБЕП-САЛДАРЛЫҚ ҚАТЫНАСТАРДЫҢ АЛҒАШҚЫ КОРПУСТЫҚ АННОТАЦИЯСЫ

Abstract

In this article, the KAZCausal corpus attempts for the first time to annotate causal relationships in modern Kazakh. The study examines the structural and methodological foundations of the KAZCausal corpus. In addition, the first attempt was made to systematically record causal and resultative relationships in texts based on scientific, journalistic, and Internet information materials, and the results of this study make a significant contribution to the field of natural language processing (NLP), integrating with the field of linguistics. In the course of the study, the process of corpus creation, the annotation scheme, and linguistic criteria will be discussed as prerequisites for creating a corpus applied to syntactic and semantic cause-and-effect configurations in the modern Kazakh language. This article will examine a number of tools, models, and corpora that have been successfully used in research on other languages for contextualization in current NLP experiments. These include the Humor (High-speed Unification Morphology) annotator for morphological processing, the POS tagging model, the Multi-tasl learning and Subdomain adaptation model, Multilingual BERT, the multilingual XLM-R model, as well as the Kazakh Language Corpus-KLC with wide coverage, Penn Discourse Treebank (PDTB), CausBank, and BECause, the main differences and innovative aspects of the KAZCausal corpus aimed at annotating causal relationships were examined, as well as features such as annotation principles, scope of application, and structural nature. In the future, the KAZCausal corpus may become the basis for application in other Turkic-language corpora, effective use of the Kazakh language in computer processing, development of semantic search technologies in information systems, and improvement of NLP models.