Experimental release: using n-grams

This release is for experimenting with the use of n-grams instead of tokens for matching. This should make the matching more robust against variations in word morphology. For example, 'infantile' will now match (partially) with 'infant', which was not the case before.