اهداف تشریح و کاربرد در جابجایی سیستم جریان کاری

Goals, which can be described as states that an agent would like to realize, is an important concept for intelligent agent systems. The representation of goals and the ability to reason about them are the major problems in goal-oriented analysis and modeling techniques, especially in intelligent agent system, as goals are more stable than other abstractions (e.g. user stories). Description Logics (DLs) is a formal tool of knowledge representation and reasoning. In this paper, we construct a framework with explicit representation and formal semantics of goals-Goal Description Logics (GDLs), which integrates two aspects of goals: declarative (a description of the state of sought), and procedural (a set of plans for achieving the goal), into one concept based on Description Logics (DLs). In addition, goals reasoning, especially goal matchmaking in GDLs, is studied using its effective judgment to concept subsumption. We propose a conceptual model of goal-based migration workflow system (GMWfS) based on GDLs, and illustrate an application. We also present preliminary experimental results on an implementation of these ideas. Compared to traditional workflow methods, GOMWfS is more flexible and intelligent

Intelligent agent, which has the ability to operate in dynamic and complex domains, has not only become one of the central topics of artificial intelligence, but also the mainstream of computer science. Many researchers take up agent-oriented programming as a new and exciting paradigm to investigate, and has significant applications in a wide range of domains (Bratman, 1987 and Rao and Georgeff, 1995). Although there has been much debate on what constitutes an agent, and which features are important, the consensus is that an intelligent agent is situated, autonomous, reactive, proactive, and social (Wooldridge, 1998). Goals, is an important concept in intelligent agent systems, which exhibit proactive behaviors. In the context of this paper, a goal is a state of the world that an agent wants to achieve. On this view, the explicit representation of goals and the ability to reason about them play an important role in agent systems. Several works have been and still are concerned with defining appropriate frameworks for advanced, intelligent, and automated problem solving using goal-driven architecture. Hindriks et al. developed an agent programming language called GOAL (Goal Oriented Agent Language) (Hindriks, de boer, van der Hoek, & Meyer, 2000). They provided a logical formalism, a programming language, and a set of formal semantics that relates the logical formalism to the programming language. In contrast to other attempts (Soham, 1993 and Wobcke, 2000), it bridges the gap between logical formalisms and a programming language that realizes the logic. Their logic requires that goals not be entailed by beliefs (i.e. they are not already achieved) and that goals be satisfiable. In PRACTIONIS, a goal is considered as an analysis, a design, and an implementation abstraction compliant to the semantics described (Morreale et al., 2006). In other words, PRACTIONIST agents can be programmed in terms of goals, which then will be related to either desires or intentions according to whether some specific conditions are satisfied or not. van Riemsdijk et al. attempt to incorporate both the declarative and procedural aspect of goals into their Dribble programming language ( van Riemsdijk, van der Hoek, & Meyer, 2003). They represent goals and beliefs as propositional formulae, and provide operational semantics for mental state changes of the agent and the execution of goals. Inspired by Dribble, Dastani et al. extended the agent programming language 3APL (An Abstract Agent Programming Language) to include the declarative aspect of goals ( Dastani et al., 2003 and Thangarajah, 2004). In the extended 3APL, goals represent situations that the agent wants to realize, and plans are introduced as procedures to allow the agent to satisfy these goals. Systems like Agent-0 (Soham, 1993), AgentSpeak(L) (Rao, 1996), and JACK (Busetta, Rönnquist, Hodgson, & Lucas, 1998), do not represent goals explicitly, but capture goals implicitly. In conclusion, all these notions are structures that built from the actions. Therefore they are similar in nature to plans. Description Logics (DLs) are a family of knowledge representation formalisms. They are based on the notion of concepts and roles, and are mainly characterized by constructors that allow complex concepts and roles to be built from atomic ones ( Horrocks and Sattler, 1999 and Baader et al., 2002). The main benefit from these knowledge languages is that sound and complete algorithms for the subsumption and satisfiability problems can be defined. A DLs reasoner solves the problems of equivalence, satisfiability and subsumption. A key difference between DLs and the standard representation formalisms based on First-Order Logic, e.g. relational and deductive databases, is that DLs provide an arena for exploring new sets of “logical connectives” – the constructors used to form composite descriptions – that are different from the standard connectives such as conjunction, universal quantifiers, etc. Therefore, DLs provide a new space in which to search for expressive and effectively computable representation languages. Moreover, although it is possible to translate many aspects of DLs currently encountered into First-Order Logic, reasoning with such a translation would be a very poor substitute because DLs-based systems reason in a way that does not resemble standard theorem proving (e.g. by making use of imperative programming features). We have presented a novel way to develop a representation of goals based on Description Logics (DLs), which are able to represent structural knowledge in a formal and well-understood way ( Xiuguo and Tongtong, 2008 and Xiuguo et al., 2008). In those papers, our aim is to allow agent implementation platforms to be more faithful to their theoretical foundations, and to provide a better handling of goals. Based on above reasons, we construct a framework with explicit representation and formal semantics of goals-Goal Description Logics (GDLs), which integrates two types of goals: declarative goals and procedural goals into one concept, using effective representation and reasoning capability of Description Logics (DLs), especially its effective judgment to concept subsumption. The main contributions of this paper include: • A logic framework to express goals description in intelligent agents, including syntax and semantics; • Reasoning, especially goal hierarchy construction algorithms in Goal Description Logic (GDLs); • A conceptual model of goal-oriented migration workflow system (GMWfS); • Evaluation of goal-based workflow system and process-based workflow system. The remainder of the paper is structured as follows: in the next section (Section 2), we define a framework of goals and plans-Goal Description Logics (GDLs). In Section 3, we present goal hierarchy algorithms, which is an important reasoning problem in GDLs. Then, in Section 4 we propose a goal-based migrating workflow model, and report experiments to show the costs and benefits of goal-based workflow system compared to traditional workflow system. Section 5 reports on background work and the related work, to finally draw some conclusions and propose some further research topics in Section 6.

Goals as used in agent programming describe situations that the agent wants to reach. The use of an explicit representation of goals provides much more flexibility in problem solving. In agent programming languages, goals are often considered in a procedural way. In most agent specification logics on the other hand, goals are employed in a declarative way. The GOAL language uses declarative goals but lacks a sufficiently powerful notation of procedural goals. Plans cannot use sequences and are limited to being reactive. A few other programming languages have also claimed to incorporate the notion of declarative goals, such as AgentSpeak(L). In this logic, the notion of goal is defined in terms of achievement-goals. In this way, it is difficult to decouple plan failure from goal failure. In this paper, we propose a framework for goal representation in agent system. Furthermore, we discussed goal matchmaking. It is only the first proposal for a formalism describing the functionality of goals, which is widely used in agent, especially in mobile agent fields. We anticipate that this work will form the basis of significant development of agent systems with explicit representation of goals, including those based on the popular BDI model, and enhancing the intelligent capabilities.