Semantic Web Agents Interoperate as Distributed (Systems) Based on Trust

Due to the dynamic nature of the multifunctional system of semantic web and existence of many agents in it, identifying reliable agents is one of the challenges for creating communication in semantic web. Assess the reliability of these systems are based on past behavior of agents. There are two main sources of information about the past behavior of agents: Direct connection between the proxy and information obtained from other agents community. Despite agents with different styles and opinions, the proposed trust model must be adjusted according to the needs of the agency. In addition for calculating the reliability of agents, should collect information about its behavior in past from agent community. Trust has also various aspects and the trust of supposed agent may be different in various dimensions. The proposed method of this paper uses a distributed model based on social networks to gather information from environment and it evaluates validity based on collected evidence by using artificial social networks. Social networks were training according to agent’s view all the time. Thus, the importance of the collected recommendations would adjust based on agent’s view of trust evaluation. Also proposed model in this paper takes advantage of the ontology to represent the trust; thereby reliability levels are calculated in different dimensions. This paper implements a context by jade firmware to evaluate the performance of the proposed model. Performance of this model is tested in practice.

Due to the dynamic nature of the multifunctional system of semantic web and existence of many agents in it, identifying reliable agents is one of the challenges for creating communication in semantic web. Assess the reliability of these systems are based on past behavior of agents. There are two main sources of information about the past behavior of agents: Direct connection between the proxy and information obtained from other agents community. Despite agents with different styles and opinions, the proposed trust model must be adjusted according to the needs of the agency. In addition for calculating the reliability of agents, should collect information about its behavior in past from agent community. Trust has also various aspects and the trust of supposed agent may be different in various dimensions. The proposed method of this paper uses a distributed model based on social networks to gather information from environment and it evaluates validity based on collected evidence by using artificial social networks. Social networks were training according to agent’s view all the time. Thus, the importance of the collected recommendations would adjust based on agent’s view of trust evaluation. Also proposed model in this paper takes advantage of the ontology to represent the trust; thereby reliability levels are calculated in different dimensions. This paper implements a context by jade firmware to evaluate the performance of the proposed model. Performance of this model is tested in practice.