Research On Realtime-task-oriented Autonomic Service Cooperation Model, Formal Semantics And Its Validation

With the science and technology developing, many distributed open complex systems have emerged on the internet, among whose operations the Real-time Task Solving (RTS) is a key problem. Service-oriented Computing (SOC) has provided some kind of more flexible and efficient solutions for RTS, and the basic idea is such to construct an instant combo service via the automatic service composition and disassemble the combo after completing the task.The SOC-based RTS has shown its deficiencies confronting the practical application scenarios which are becoming more and more complicated.It is difficult for SOC to get a good balance between the efficiency and precision of automatization, which baffles the applicability of RTS. The solving participants can't handle the dynamic and unpredictable internet environment and requirements for lacking intelligence and rationality. A top-level control mechanism must be utilized to deal with emergent events through a global viewpoint of the open complex system, which compels the systems back into a closed state. Besides, there is also a need to formally analyze and validate the RTS execution process to ensure the expected goal can be attained. The current methodology of validation is limited to do that only to check whether the combo service operated well and whether the users'goal is achieved. Even if the validating is complete with the right result, it is hard to guarantee the RTS solution will function in gear and hard to settle the problem of participant escaping. This thesis aims at the problems above, and proposes novel solutions which are listed below.Firstly, to promote the extent of automatization, a integrated method of improving the performance is put forward from the point of service description, service discovery and service composition respectively. On the part of service description, a methodology of depicting services'semantics in a top-down manner via tri-level is presented. The three levels are application level, which provides a model of service semantics knowledge acquiring on the basis of ontology, semantic level, which supports a multi-attribute description, and implementation level, which encapsulates the semantics with the extended task circumstance description pattern. It provides the service developers with the flexibility and facility to create ontology quickly and conveniently for different application domains. An application domain professional oriented ontology-creating tool is also introduced. The GUI of this tool can be used as a wizard guiding the users creating service description. This tool promotes the extent of automatization of describing service by weakening the dependency of ontology creators on the IT knowledge. On the part of service discovery, a mechanism of combining the multilevel lexical hashing with semantic matchmaking is put forward. This mechanism breaks the service description model into blocks of different granular, and hashing functions are designed with respect to every granular block. The service discovery process is therefore transformed into the multilevel lexical hashing by the ascending granular order, and the collision of hashing is resolved by the subsumption based semantic matchmaking. On the part of service composition, a DCWMG based dynamic service composition algorithm is proposed regarding to the service repository set up with the service discovery mechanism above. This algorithm integrates the depth-first search with the breadth-first search, defining the reverse edge via coloring technique, and provides a linear time searching of all connected paths between two nodes.Secondly, how to make computing entities intelligent for RTS is introduced. An external governing mechanism internalization model is presented. This model imports the commitment semamtics into the Cohen&Levesque BDI model, and fusing the commitments&conventions cooperation model by N.R.Jennings. This internalization model compiles the external macro-governing mechanism into the internal belief, desire and intention of computing entities via commitment semantics, and provides a foundation for computing entities to abide by the application layer instructions.Thirdly, some attention is also paid to both the formal semantics of the RTS process and its validating methods. A coarse granular ATCL logic is constructed for the commitment based cooperation model which abstracts the one depicted above. This logic extends the cooperating operator《》into《C:ζ:ω:Θ》via commitment semantics denoting the organization C commits starting cooperating according to planΘas soon as the conditionsξ, is satisfied, and promises attaining some negotiated goals beforeωbecomes true. An action based transition system is presented at first to model the agent organization founded on the basis of the conditional commitment carrying with deadline. The syntax and semantics of ATCL are offered, whose expression capability is exemplified. The model checking algorithm and its complexity are also introduced.