US8095530B1
patent
A computer-implemented method includes receiving a plu rality of character strings. The number of strings (M) in the plurality of strings having a unique Substring of X characters at an extremity of the string is determined, the number of strings (N) in the plurality of strings having at least X char acters in the string is determined. A probability is determined, based on a predetermined model for a distribution of charac ters in the strings, that the unique Substring of X characters would occur M or more times out of the N strings, given that the unique character string occurs at least once. Based on the probability, the number M, and the number N, it is determined that the unique character string is a significant affix in the plurality of character strings, and the unique character string
25 claims. 92,861 characters of text.