Extração de informação apoiada por uma ontologia de domínio da enfermagem baseada em evidências

Abstract

The growing volume of documents with relevant information available in electronic media is a strategic resource for the health professional that uses these sources to support decision making. Moreover, it is not feasible to analyze a significant amount of documents in a short time without the aid of a computational tool. The extraction of information aims to extract from textual documents only the relevants informations defined by the user. This information is mapped using techniques of text classification. These techniques are based on information contained in a formal style. Regarding the health area, research shows that professionals decision making centred on domain evidence-based documents of updated content. In this context, the domain of Evidence-Based Nursing (EBN) comes to the practical application of the hold decision nurses process information. However, Ontologies are focused on formal and explicit representation of a shared conceptualization. The use of ontologies in this work is related to the representation of the evidence based nursing domain that has not been formally represented in a computational structure, as well as serve as a support for text classification technique for extracting information from scientific documents in nursing. Thus, the purpose of this study is to design an extractor mechanism of information that combines a hybrid form of information extraction with a domain ontology for evidence-based nursing. The mechanism serves to assist in the extraction, storage and reuse of relevant information in order to optimize decision-making of nurses. To validate this work, with respect to the extractor mechanism, classification and extraction techniques have been applied to documents of the National Institute of Health research (NHS), through a case study. After the construction of the domain ontology, it has been validated through an experimental study.