Describir: An effective named entity similarity metric for comparing data from multiple sources with varying syntax.