Complex Decision-making Task Solving Based On Ontology: Representation Method And Key Technologies

Decision-making task solving is a hot issue in the field of decision science and AI. At present, the theory and methodology research of decision-making solving is mainly based on Decision Support System (DSS). However, most of the existing DSS cannot meet the demand for fast decision making, intelligent decision making and complex decision making, because the architecture and technologies of them cannot availably support the functions needed for complex decision-making such as knowledge management and automatic model matchmaking and composition, and not well deal with the role division of human and computer. In this paper, a series of key technologies are studied in the framework of Web Services packaging technology, Formal Concept Lattice, graph search and services scientific research. Ontology, fuzzy logic and Bayesian network are introduced to discuss the representation of complex and uncertain knowledge. The main contribution of this paper is summarized as follows.1. An ontology-based formal method of complex decision tasks is proposed in the paper, and the general process of decision-making tasks of complex problem solving is discussed together with the formalization of the complex decision task. The methods to construct the related ontology are presented, including domain knowledge ontology, decision model ontology as well as decision task ontology. In addition, the hierarchical relationships among various ontologism and mechanism are analyzed.2. For the fuzziness of human thought as well as language, the fuzzy logic method is introduced into the ontology so as to realize the fuzzy extension of the formalization for complex decision task. Fuzzy theory is introduced into OWL2 and fuzzy description logic is presented with the definition of syntax and semantics, and the crisp representation is applied so that existing reasoning mechanisms are available, upon which the fuzzy extension of related decision task ontology is realized. This fuzzy extension is used to calculate the text similarity in service matchmaking.3. For the stochastic and probabilistic of events, the Bayesian network is introduced into ontology so as to realize the probabilistic extension of the formalization for complex decision task. Probabilistic reasoning in the Bayesian network is introduced into OWL2 and probabilistic description logic is presented with the definition of syntax and semantics, and the ontology-based description are transferred to Bayesian network-based description, upon which the probabilistic extension of related decision task ontology is realized.4. A decomposition method of complex decision task based on improved HTN planning is proposed. With the redefinitions of the operator and method of the HTN planning and the domain- dependent extension of the conventional HTN planning algorithm, the task domain knowledge is introduced, and the complex constructions such as parallel, repeat, choice, are describable, upon which the automatic decomposition of domain-dependent decision task is realized.5. A human-computer interaction coordination mechanism of decision tasks solving based on extended WS-BPEL is proposed. An abstract workflow is described using the extended WS-BPEL, and a concept named personware service is introduced to deal with the manual participation as services, and the action states and transformation mechanisms of the dispatcher are defined to realize the human-computer coordination, upon which the human beings as business process participants are coordinated together with the Web services.6. The prototype of the Service-oriented Integrative Decision Support Platform(IDSP-S) is designed and realized, and the case study validates the feasibility and validity of the methods and technologies in this paper.