Research On Intelligent Agent For End-user In Cloud Computing

In the computer field, the high degree of intelligence and automation has always been a dream of human benings. From artificial intelligence wave in the1980s, humans have hoped that machines can help people deal with some of the things, instead of a part of human ability. Therefore, a concept, agents, was proposed with the development of sensors, software engineering, network technology and artificial intelligence.In recent years, with the progress of the information society, a variety of information resources have widely emerged. For example, the development of the Internet of Things provides perception information in all aspects of human life. With the emergence of Web service, application development become easier and more diverse than before. The rapid development of mobile Internet provides a higher-speed channel for end users to get more information. The birth and development of cloud computing technology allow end users can obtain massive computing resources and storage resources. It will provide convenience and help for the production and life of human, if agents can take advantage of these resources effectively.In this dissertation, cloud computing, networking and service-oriented computing is involved to research an intelligent agent for end-users. We mainly carried out the following research:Firstly, we proposed a concept of intelligent agents for end users in cloud computing environments. We called this concept "Cloud-Brain". Through the Cloud-Brain, end users can upload a part of their knowledge by the form of facts and rules in the cloud environment. Then this knowledge combined with a wide range of sensory data in the inference engine. The Cloud-Brain will provide appropriate function to users in virtue of service-oriented approach. The intelligent agent simulate logic of users in the cloud environment instead of part of the user’s thinking, and act to the environment around users on the results of thinking.Secondly, for the need of integration of a variety of data resources from Internet in Cloud-Brain, the integration of heterogeneous data from different sources has been studied. In this dissertation, an ontology-based fact model, a data source model and some mapping strategies are proposed. By building these data model in the Cloud-Brain, the heterogeneous data can be perceived effectively in the system.Thirdly, the Cloud-Brain model exists interaction issues in services and human-machine. Therefore, the interaction model between entities is studied. A uniform rule-based interaction model has been proposed. The takes advantage of the characteristics of the Cloud-Brain model. In this interaction model, we deal with the various interaction problem between entities in the form of facts and rules. And we have achieved the interaction service and client to verify the interaction model.Fourthly, in order to improve the capability and scalability of rule engine in the cloud computing environment, we proposed and implemented a rules engine under the Actor model based on cloud computing environment. In the rules engine, the Rete matching algorithm and Actor model was involved to distribute the processes. The entire Rete network of rule engine can be assigned to several virtual machines. The rules engine can take full advantage of in the cloud computing resources environment according to the scale. It can handle large-scale rules and facts. On this basis, we proposed a resource model and process allocation algorithm to optimize the system performance in cloud computing environment.Finally, in order to verify the feasibility and performance of Cloud-Brain intelligent agent, we have implemented a prototype system and deployed the distributed system in cloud environment. Based on this system, we verified the function of Cloud-Brain intelligent agent and implemented an example application scenario with campus environment.This dissertation provides an intelligent agent, which can deal with the perception information for end users based on cloud environment. The facts model of the intelligent agent, interaction model, and rules engine in cloud are discussed in the dissertation.