The data representations of the chromosomes that will represent the candidate DFA were introduced in 3.2. To comply with that specification the suffixes (tokens) tabulated in the previous subsection are enumerated. So the count of genes is 14 and the allele of all genes are . The fact that all states are also final states is reflected in the fitness function of the GA.
The production of the testbed is automated. The testbed consists of strings of numbers. Each number in a string corresponds to a noun inflection suffix. In this way the whole string represents a valid noun formation. A random string generator is implemented which takes as input a DFA and produces random strings accepted by that DFA and are of various lengths that vary from a minimal length to a maximal length. The uniqueness of the random strings is ensured. The testbed generator is fed with the above pictured noun inflection DFA and asked to produce strings that correspond to valid noun instances. The run was performed with a testbed of 500 strings where the lengths varied in the range [5,15].