The method of second differences in parallel algorithms of linguistic analysis

A new algorithm is proposed for approximating a matrix by overlapping blocks. The algorithm saves much computer time and gives a better approximation of empirical data that conventional methods do.