Machine learning and word sense disambiguation in the biomedical domain: design and evaluation issues