| (1) The number of individual ants is ; maximum number of iterations is ; side length of localized area |
| is ; velocity of ants, , takes a random number. |
| (2) The data object is projected onto a plane and each object is assigned a point (). |
| (3) Each ant is initialized to unload and randomly selects an object. |
| (4) for |
| for |
| Calculate the average similarity of object. |
| If the antis unloaded, then calculate the probability of picking up, . If is greater than a random |
| probability, while on the same time the object is not being picked up by other ants, the ant picks up the |
| object and moves elsewhere randomly, and marks that it is load; otherwise, the ant refuses to pick up the |
| object and selects other objects randomly. |
| If the antis loaded, then calculate the probability of putting down, . If is greater than a random |
| probability, the ant puts down the object and marks that it is unload, and then re-select a new |
| object. Otherwise, the ant continues to move the object to a new location. |
| End |
| End |
| (5) for |
| If an object is isolated or the number of its neighborhood is less than a certain constant, then mark the |
| object as an isolated point; otherwise, the object is assigned to a cluster serial number, and recursively their |
| neighborhood object tag for the same serial number. |
| End |
