Research On Process Ontology Based Heterogeneous Agent Cooperation Technology

Recently, the flexible and complex applicationsacross multi-organizations in opennetwork environment have become the common requirement in many fields. With theimprovement of Agent and Multi-Agent System(MAS) technology, it more clearly isconsidered that they are the suitable paradigm to build such complex and distributedapplications.In thismetaphor, complex system functions are realized bycooperation ofthe heterogeneous agents which belong to different organizations. So, it is veryimportant to research the cooperation of heterogeneous agents in the open dynamicnetworkenvironment.The traditional agent cooperation methods always share the process frameworkimplicitly defined by system designers to guarantee the coherence of computing, but itis not suitable inthe open dynamicenvironment. The Ontologytechnology supports thecooperation of heterogeneous agents in the open dynamic environment by facilitatingthe communication of agents and the matchmaking of services. It does not make use ofthe Ontology technology sufficiently, and can't improve the efficiency of the agents'cooperation. So, this dissertation propose a Process Ontology which can support thecooperation of heterogeneous agents, and thecooperation mechanism of heterogeneousagentsbasedonProcessOntology.Theprimaryworkandmaincontributionsofthisdissertationinclude:1. The research on Process Ontology which can support the cooperation ofheterogeneousagents.To provide sharing semantics for the cooperation of heterogeneous agents, thisdissertation makes an analysis on several process description, representation andspecification language firstly, and extracts the core modeling elements and logicrelationships between them (which is called DL-P Core); Secondly, Process Ontologyspecification DL-P (Description Logics-Process), which can represent the complexprocess, is developed based on the DL-P Core. This specification is based onDLs(Description Logics), so that, it is compatible with the current ontology languagestandard of the semantic web, and can be published on the web. Such that, it can beaccessedandusedbyheterogeneous agents. Then, this dissertationpresents amethodtobuild Process Ontology based on DL-P, and studies the basic Axioms and the basicreasoning tasks of Process Ontology Knowledge Base. At last, an algorithm to detecttheprocessmodelbasedonDL-Pisproposed.2.AcooperationmethodoftheheterogeneousagentsbasedonProcessOntology.According the research of Jennings, Woodridge and other researchers, thisdissertationadopts theCommitment/Conventions andJointCommitment/Joint Intentiontobringoutaformal frameworkofagentcooperationandanagents cooperationprocess based on this framework. The key issues which support the agent cooperation areanalyzed theoretically, and the supporting technologies of the heterogeneous agentcooperation in the open dynamic environment are pointed out. Based on these works,the heterogeneous agent cooperation architecture in the open dynamic environment isdesigned, and the heterogeneous agents cooperation space & management technologyalso studied based on Process Ontology. At last, this dissertation studies how to useProcess Ontologyto facilitate the heterogeneous agent cooperation, which includes thebuilding, executing and monitoring of cooperation based on the cooperation space andcooperation management technology, and a new heterogeneous agent cooperationmechanismintheopendynamicnetworkenvironment namedCoMBPOispresented.3.CoMBPOorientedagentdesignandimplementationtechnology.Based on the BDI agent model, an agent design modeling method using ProcessOntology is proposed. And a multi-thread implementation method of BDI agent ispresented. The agent design modeling method matches Process Ontology with thedevelopment of agents. So, it can reducethe difficulty of the development of agents byanalyzing and designing the agent's mental elements(Belief, Goal and Plan, etc.) basedProcess Ontology and avoid the disagreement between knowledge engineers andprogrammers. An AGENTFRAME structure which can express the belief, desire andintention explicitly is presented as the multi-thread implementation method of BDIagent. And the MAS development platform called MBOS, which has used theAGENTFRAME as its'agent kernel, has been tested and evaluated by severalapplication instances. The result shows that AGENTFRAME is a feasible agentarchitecture, and the agent based on this kind of architecture is able to runpersistentlyandaccomplish manyintelligentbehaviorssuchasreasoningandcooperation.Then thisdissertation presents a method to design and develop a BDI agent based on ProcessOntology, It is about how to design and develop the Belief, Goal, Capabilityand Plan,andhowtoexecutethePlan.4. Finally, on the military application background of the Oil Logistics Networks,an Agent Based Oil Logistics Networks system(ABOLN) has been developed to testand prove the feasibility of the technology presented above. At first, A ProcessOntology about the Oil Logistics Networks has been built based on DL-P, and theagents have been developed according to the CoMBPO oriented agent design andimplementation method CoMBPO oriented and the multi-thread implementationmethod of BDI agent; when the prototype running, the cooperation method of theheterogeneous agents based on Process Ontology has been tested: from the agent'sregister, cooperation launching, finding cooperative partners to implementation of thecooperation. These agents cooperate with the other agents to guarantee the oil logistics,even at emergency situation. Thisexperiment shows the research achievement above isvaliditytofacilitatethecooperationofheterogeneousagent. To sum up, to facilitate the cooperation of heterogeneous agents, this dissertationpresents the general Process Ontology DL-P which can support the cooperation ofheterogeneousagents in the open dynamic environment, anew cooperation mechanismof heterogeneous agents based on Process Ontology, and the agent design andimplementation technology based on Process Ontology. All of these achievementsexhibitanewtechnologyapproachtosolve the puzzles in cooperationofheterogeneousagent to improve the efficiency of agent cooperation, which make a nice groundworkforfuture research and engineeringoncooperation andintegrationof the heterogeneoussystemsintheopendynamicenvironment.