Bio- Inspired Algorithm for Text Classification: Comparison Between Ant Colony Algorithm and Olex-GA

Artificial Ant Colony Optimization (ACO) is a probabilistic technique which has a reputation to solve travelling salesman problem (TSP) in an efficient way. The use of pheromone is the distinctive criteria between ACO and other techniques. To its use, pheromone concentration is adjusted depending on the solutions that has been discovered while randomly attempt in choosing path from nodes. This research concentrates on two parts, pheromone usage and adaptation of ACO in classifying text document. The research intended to keep standard setting untouched except the vertices and the node calculation part where they are necessary to adapt ACO for text document classification. The idea of the combination is used to experiment shortest path relationship to text document problem. It was found that the result is comparable to a wrapper method, Olex-Genetic Algorithm (GA). ACO is found to have 7.25% higher average accuracy than GA described with supporting experiments. For future research, the ACO as used in solving TSP is hoped to be further enhanced especially classification accuracy.