Methods for Recognizing Nested Terms

Terms are defined as words or phrases that denote concepts of a specific domain, and knowing them is important for domain analysis, machine translation, or domain-specific information retrieval. V arious approaches have been proposed for automatic term extraction. However, automatic methods do not yet achieve the quality of manual term analysis. During recent years, machine learning methods have been intensively studied (Loukachevitch, 2012; Charalampakis et al., 2016; Nadif and Role, 2021). The application of machine learning improves the quality of term extraction, but requires creating training datasets. In addition, the transfer of a trained model from one domain to another usually leads to degradation of the performance of term extraction. Currently, language models (Xie et al., 2022; Liu et al., 2020) are texted in automatic term extraction.