Methods for automatic term recognition in domain-specific text collections: A survey

Applications related to domain specific text processing often use glossaries and ontologies, and the main step of such resource construction is term recognition. This paper presents a survey of existing definitions of the term and its linguistic features, formulates the task definition for term recognition, and analyzes presently-available methods for automatic term recognition, such as methods for candidates collection, methods based on statistics and contexts of term occurrences, methods using topic models, and methods based on external resources (such as text collections from other domains, ontologies, and Wikipedia). This paper also provides an overview of standard methodologies and datasets for experimental research.