Evolving ontologies using an adaptive multi-agent system based on ontologist-feedback

In a changing environment, it is necessary to update the ontology to new knowledge and user needs. However, ontology evolution is still a time-consuming and complex task. In this paper we propose an extended approach of an earlier work in ontology evolution based on an adaptive multi-agent system (AMAS). In fact, we seek to personalize the results proposed by the AMAS to the ontologist-feedback. First, we enhance the agents with an adaptive behavior enabling them to react to the ontologist's feedback. The ontologist gives his/her action (elementary and composite changes) towards the AMAS proposals. He/She can also add new terms and concepts. Then, the AMAS reacts and self-organizes to produce an updated ontology with new proposals. This process is repeated until a satisfactory state of the ontology is obtained. The experiments prove that the adaptive skills we added to agents help them to detect the uselessness of some proposals, to avoid the useless and wrong ones and to propose others.