In this paper, we present Dynamic Terminology Enhancement Method (DTEM) to support enrichment and extensibility in a biosignal integration system called ROISES (Research Oriented Integration System for ECG signals), which integrates diversely encoded ECG signals and the corresponding annotation and metadata. The diverse datasources are homogenized through the mapping of their schemas to an ECG specialized global ontology (GO). DTE method combines UMLS rich terminology and machine learning techniques to first determine the suitability of a term to constitute global ontology's class and secondly locate its position in GO's hierarchy.