Named Entity Extraction in Requirement Specification: A Comparison
Resumen
Software requirements specifications generally are written in natural language. Identifying and extracting the main concepts involved in a requirements specification could be useful for the development process, quality assurance, and software maintenance. However, a computer agent is not able to process and understand immediately the content and information included in the natural language documents. Named entity extraction is a task that involves recognizing entities in a text and linking them to a knowledge graph to disambiguate them. In the field of requirements, applying this task can be useful for building structures that allow for the representation and efficient management of complex information. Different tools are focused in entity extraction and then an entity linking with a specific knowledge graph such as Wikidata. This work compares different named entity extraction tools in the task of extracting entities in a requirements specification.