UCM at CLEF eHealth 2013 Shared Task1b

We are developing a system that analyze medical reports and extract a SNOMED-CT based concept representation. The more interesting characteristic of our system is not only that it can detect the concepts. It also takes into account if they appear in an affirmative, negative or speculative context. The system also separates the concept representation according to the structure of the document. Our system takes these steps: automatic orthographic correction, acronyms and abbreviation detection, negation and speculation phrase detection and medical concepts detection. For participating in Task 1 we have adapted our system in order to obtain the mentions that belong to the Disorders UMLS semantic group. The approach is based on using MetaMap to detect the concepts and the spans. Our aim was to identify what was the best way to use MetaMap in our system to solve the Task 1.