Stores the first layer, i.e., the entry points in the trie
Generates BIO labels for this sequence when complete trie paths match When multiple paths match, the longest one is kept
Returns the length of the matched span, or -1 if nothing matched When multiple paths are found, the longest one is kept Text must be normalized (i.e., case folding) BEFORE this call, if necessary!
For stats
A more efficient trie implementation, where the first layer is stored as a hash map for efficiency (the rest are the usual trees) The find* methods are thread safe User: mihais Date: 5/12/15