Ji Gao

Institution-governed cross-domain agent service cooperation: a model for trusted and autonomic service cooperation

Constructing Virtual Organizations (VOs) by creating service cooperation (i.e. service-oriented cooperation) has become a mainstream approach for reforming the development of application software systems in Web environments. However, the inherent non-controllability of business services across different management domains has brought on the so-called “trust” crisis that the success and benefit of cooperation cannot be ensured. It is this crisis that cumbers the achievement of autonomic cooperation and thereby the large-scale deployment of VOs. Therefore, this paper proposes a model, called IGTASC, to conquer this crisis and make service cooperation both trusted and autonomic by developing three closely-coupled technologies: institution-governed cooperation, policy-driven self-management, and community facilitation management.

A Hybrid E-institution Model: From the Static to the Dynamic

The formation and operation of the virtual organizations (VO) is to to facilitate resource sharing and problem solving among software and/or human agents. However, due to the selfishness and autonomy of agents, VOs are confronted with the credibility crisis. E-institutions are perfict means to sovle the problems of VO credibility crisis. But how E-institutions facilitate the VO in different stages and what is the E-institutions architecture is still blurred. In this paper, we propose a dynamic hybrid E-institution model to provide assistance to the agent interactions. First, the E-institutions are modelled using static elements to make the agent cooperation trustable and predictable; second, the E-institutions are aslo modelled using dynamic elements, instantiated information; third, the E-institutions are modelled using conditions under which the static elements are translated into instantiated (dynamic) ones. Different conditions serve different E-institution goals and thus make the E-institution dynamic. This paper presents explicitly formal descriptions of the static organization structure, dynamic information and institutional services to different VO stages.

Task Oriented Web Service Requirement Framework

Aiming to overcome the mismatch between coarse grain of Web services functions and fine grain of current Web service requirements with inputs and outputs as functional requirements, a three layer task oriented Web service requirement framework is proposed. The top the is task layer which defines the tasks and their relationships, the middle is the method layer which represents multiple implementations of the primitive task, and the low is the operation layer which represents the abstract features of typical operations that a method owns. The task oriented Web service requirement framework proposed well matches the coarse grain feature of Web services functions and provides a more natural way to express the service requirements.

SNMG: a social-level norm-based methodology for macro-governing service collaboration processes

In order to adapt to the accelerative open tendency of collaborations between enterprises, this paper proposes a Social-level Norm-based methodology for Macro-Governing service collaboration processes, called SNMG, to regulate and control the social-level visible macro-behaviors of the social individuals participating in collaborations. SNMG not only can remove effectively the uncontrollability hindrance confronted with by open social activities, but also enables across-management-domain collaborations to be implemented by uniting the centralized controls of social individuals for respective social activities. Therefore, this paper provides a brand-new system construction mode to promote the development and large-scale deployment of service collaborations.

SMGSC: Social-Level Macro-Governing Methodology for Cross-Management-Domain Service Collaboration Processes

In order to adapt to the accelerative open tendency of collaborations between enterprises, this paper proposes the cross-management-domain social-level macro-governing mode for regulating and controlling the social-level visible macro-behaviors of the social individuals participating in collaborations. Then this mode is achieved by creating Social-level Macro-Governing methodology for cross-management-domain Service Collaboration processes, called SMGSC, which consists of two levels of standalone and complementary technologies: the social -- level collaboration process norm system represented as social dependence norm sets and the rational agents whose macro-behaviors in collaboration processes conform to norms. Since the rational agents forming dynamically collaboration relationships can make their macro-behaviors governed by social dependence norms, SMGSC not only can remove effectively the uncontrollability hindrance confronted with by open social activities, but also enables cross-management-domain collaborations to be implemented by uniting the centralized controls of social individuals for respective social activities. Therefore, this paper provides a brand-new system construction mode to promote the convenient development and large-scale deployment of service collaborations.

Social dependence evolution networks for governing agent service cooperation processes

In order to remove the hindrance of uncontrollability for cooperation processes which the service cooperation across management domains is faced with, this paper proposes to use software agents as the brokers for enterprises to provide / consume business services through Internet, and to develop the social-level norm system based on social dependence evolution networks for governing agent service cooperation processes. Therefore, cooperation processes can not only be controlled reliably and coherently, but also acquire the robust and efficient self-adaptive capability by configuring controllable abnormity transaction activities across management domains.

Active mechanism of deceit detection for multi-agent based interaction

Unlike traditional detection methods, which are essentially a passive approach, this paper presents a novel approach active deceit detection (ADD) against the deception problem in multi-agent interaction. ADD does not only receive and process the communicative messages, but also proactively interact to environment and interacted agents via negotiation to explore evidence for further detection. It makes detection become an active process to ensure secure collaborations in open environment. This paper presents the definition of the active deceit detection model and gives a theoretical analysis of its property on it. Moreover, an ideal estimate model of ADD, called ideal oscillation model and throw-stone test, is also further discussed.