Probabilistically split concatenated words using NLP
based on English Wikipedia uni-gram frequencies.