Domain specific named entity extraction for modeling and populating ontologies

Abstract

Automatic extraction of knowledge in modeling/enriching ontologies for domain specific applications play key role owing to the huge amount of data available in the form of documents. As manually extracting information is a tedious task, there is a need for automating this process. Use of automatic information extraction processes not only reduce the time, but also retrieves the information in a useful format. This paper proposes the use of parts of speech (POS) tagging, a Natural Language Processing (NLP) task, to group the words or entities in a text into pre-defined domain specific concepts. For the purpose of extraction, the domain concepts from available Engineering Ontology related to mechanical domain from the literature is considered. The methodology involves, parsing the text for POS tagging and then analyzing it, for grouping them into specific categories such as device, material and so on. Data required for automatic extraction is taken from various online sources describing the mechanical components, the material and process used for manufacturing those. As a start in using NLP techniques, automatic extraction of four domain concepts, device, material and process is addressed and the benefit of using it in automatic extraction of the conceptual information corresponding to an ontology is presented.