Approaches, tools, algorithms, and methods for automatic term extraction: A systematic literature mapping

Abstract

Abstract Automatic term extraction is a branch of Natural Language Processing (NLP) used to automatically generate lexicographic materials, such as glossaries, vocabularies, and dictionaries. It allows the creation of standard bases for building unified theories and translations between languages. Scientific literature shows great interest in the construction of automatic term extractors and includes several approaches, tools, algorithms, and methods that can be used for their construction; however, the number of articles in specialized databases is vast, and literature reviews are not recent. This paper presents a systematic literature mapping of the existing material for developing automatic term extractors to provide an overview of approaches, tools, algorithms, and methods used to create them. For this purpose, scientific articles in the domain published between 2015 and 2022 are reviewed and categorized. The mapping results show that among the most used approaches are statistical, with 21.85%; linguistic, with 9.75%; and hybrid, with 68.29%. In addition, there are various computational tools for terminology extraction where authors use different methods for their construction and whose results are measured under the criteria of precision and recall. Finally, 113 documents were selected to answer the research questions and to demonstrate how automatic term extractors are constructed. This paper presents a global summary of primary studies as an essential tool to approach this type of computational system construction.