Semantic Patterns Generation Using Upper Ontology

Semantic pattern (SP) generation refers to the process of identifying and understanding meaningful patterns in natural language text. Recognizing SPs is essential for a variety of natural language processing (NLP) tasks and applications, including machine translation (MT), text interpretation, text summarization, information retrieval (IR), knowledge representation, and other similar tasks. Numerous strategies are employed in the recognition of SPs, such as rule-based approaches, statistical approaches, machine learning approaches, semantic role labeling, etc. Most of these approaches necessitate large volumes of labeled training data while also being computationally demanding, particularly in deep learning models. The present research proposes a new computationally light-weight approach for automatically generating SPs in the healthcare domain using the Upper Ontology (UO) without the need for annotated datasets. The suggested approach starts by extracting the question part from the text and then takes advantage of combining and employing various name entity recognition (NERs), including the NLP toolkit’s NER. The proposed approach in this study provides a well-defined generated SPs set in the domain of healthcare, specifically patient questions. Using SPs is very important in NLP as general and in healthcare, especially. A single SP can generate a large number of patients’ questions in healthcare domain. A single SP may exactly match more than 110 questions in the dataset and, on average, match 5–17 questions.