Convert Sentence to N-dimensional point

The main problem that I have encountered in a computational linguistic is conversion of a sentence to a number representation. Each sentence is a separate unit (just like every word in it), and it has to be represented "on its own", that is, without a comparison to other sentences.

This program approximates the n-dimensional representation of a sentence as an n-dimensional point. For the simplicity's sake, n can be an integer from 1-100

Input is a sentence as a String, number of dimensions (1-100) as Integers, and Output is an array of Doubles (each axis ranges from 0-1).

The main problem of this method is that precision has to be tuned. That is, at higher number dimensions, the precision of an approximation becomes lower, while at the lower number of dimensions, precision becomes more robust. As a comparison with analogy, take a digital photograph. At lower number of dimensions photo becomes pixelated, while at the higher number of dimensions photo becomes very noisy...