Study on Ontology Partition Based on Ant Colony Algorithm

Ontology is an effective method to express knowledge. Some ontologies are large usually. The large scale of ontology brings a new problem in maintenance, release etc. Ontology can be divided into some small ontologies through ontology partition. It leads to easy use of ontology. Ontology partition method based on ant colony algorithm is presented in the paper. Not only hierarchy relations, but also the other semantic relations are considered to compute concept relativity in the method. Ontology concept hierarchy relation is taken as directed acyclic graph and common disjunctive ancestor is used to calculate concept hierarchy relativity. Ant colony algorithm is adopted to solve ontology partition. Solution construction method and local search method special to ontology partition are given in the paper. Pheromone is laid on concept and applied to construct solutions in the algorithm. Local search is put forward to improve the quality of solutions. Pheromone matrix is updated according to the quality of solutions. The results of experiments shows the method proposed in the paper gains better partition result than the other methods.