Results for other-segmenters/br-goldwater-bigrams.txt: (ignoring first 1000 utterances) ============================ Found Words ----------- Over-segmented section of a true word: 4149 (14.44%) Contained multiple whole true words: 3304 (11.50%) Contained partial true words: 509 (1.77%) Over-segmented vowel from a true word: 258 (0.90%) Over-segmented affixes from true words: 800 (2.78%) Contained an under-segmented determiner: 489 (1.70%) Contained an under-segmented collocation: 2917 (10.15%) Total: 28730 True Words ---------- Found just over-segmented: 1711 (5.67%) Found just under-segmented: 7171 (23.77%) Found both over- and under-segmented: 517 (1.71%) Found with over-segmented vowels: 258 (0.86%) Found with over-segmented affixes: 780 (2.59%) Determiners that were found under-segmented: 489 (1.62%) Total: 30167 Word Stats ---------- Correct (true pos.): 20768 Incorrect found words (false pos.): 7962 Missing true words (false neg.): 9399 Precision: 72.29% Recall: 68.84% F: 70.52% Boundary Stats -------------- Precision: 87.76% Recall: 81.86% F: 84.71% Lexicon Stats ------------- Precision: 60.80% Recall: 51.46% F: 55.75%