2014iii Autorizo a reprodução e divulgação total ou parcial deste trabalho, por qualquer meio convencional ou eletrônico, para fins de estudo e pesquisa, desde que citada a fonte. This Doctoral Thesis aim to investigate the application of text mining for the construction of a terminology that meets the procedures laid down for standardization, national and international, regarding the establishment of an Electronic Health Record (RES). International standards studied in this work were ISO 13606-1 and ISO TS 17117. The ISO 13606-1 international standard specifies the reference models for the construction of archetypes, which is the basis of the RES structure. The technical specification ISO 17117 provides the formatting of controlled terminology for the scope of health informatics. Locally, the paper analyzed the technical ABNT / ISO TR 20514 report, which gives the definition, scope and context for the RES and technical ABNT / ISO TR 12309 report aimed at ensuring the development of standardized terminology for the health sector. Several scientific studies have shown that for the construction of RES based on archetypes, researchers use market terminology such as SNOMED CT and SNOMED RT. In Brazil, there is no terminology officially developed regionally or translated into Portuguese-Brazilian who support the creation of reference models. This situation impedes the deployment of national and international standards of standardization mentioned above. In this environment, the thesis presented here built an ontology in the field of specialty of Radiology and Diagnostic Imaging based on the application of text mining methods to make efficient and effective terminology that meets the demonstrated shortcomings. The application of text mining method was performed on a sample of 2,566,358 of subject-report, consisting of subject-reports of examinations MRI, X-ray, CT and ultrasound of human anatomical regions. Based on this extraction was built an ontology containing 5,859 individuals terms, axioms 20,994 and 15,084 logical axioms. This ontology was developed using Protégé OWL language software. From the formalization of the ontology (terminology) were built Archetype Definition Language (ADL) for INSTRUCTION component for imaging examination, and ADL for COMPOSITION component of CT cervical spine, MRI Cervical and Thoracic and MRI Carotid. The study showed the applicability of text mining to generate terminology that supported the creation of ADL as recommended by rules in the IT sector in health.