Next: Mutation Up: Genetic Algorithm Previous: Genetic Algorithm

Crossover

For the purposes of this section, assume that the fields in every profile are identically ordered. The exact order is not important, as long as it is the same in all profiles. So, for example, in every profile, the newsgroup field is the first, the keyword field second, and so on.

Let and be the parent profiles: The crossover operator takes two parents and produces two offsprings. Each offspring inherits some attributes from one parent and the rest from the other. A two-point crossover is used by the crossover operator. Two points are randomly selected in the list of fields. All the fields lying between the two points are exchanged between the two parents to create two new offsprings. Since the fields in both the parents are in the same order, each offspring will have one field of each kind. The fitness of the offsprings is set to a default initial value. Formally:

Equation defines the crossover operation, where parents and produce offsprings and . The two crossover points are generated as shown in equation , where the random number function gives an integer in range specified by the two parameters inclusive (the index of the first field is 0). If the offspring profiles are as defined in equation , each field is derived as shown in equations and . The fitness of the offspring profile is set to its default value, as in equation .


MIT Media Lab - Autonomous Agents Group - agentmaster@media.mit.edu