Synthesis of Agent Behaviors and Web Services for the Semantic Web

Paul Buhler and José M. Vidal

University of South Carolina


Today' software systems are becoming more net-centric, distributed, and heterogeneous. Moore's law (processor power), Gilder's law (bandwidth expansion), and Metcalfe's law (network dynamics) predict a future that will require us to change our current perceptions about computing. Hardware, software and networking technology will combine in a milieu in which they become ubiquitous and inseparable. The acceleration of technology and time-to-market pressures make it increasingly difficult to produce software. In order to achieve the promise of the information age, software developers will require new abstractions that will allow them to manage the overwhelming complexity of this digital landscape.

Software development needs to progress from handcrafted, line-at-a-time techniques to methodologies that support reuse of existing software assets. In other words, software development needs to shift from paradigms that are purely creational to others that support compositional approaches. This change is not subtle, the current set of recommended activities and processes for the engineering of software are inadequate for the development of software using compositional techniques .

We are working to develop a component-based agent architecture that will enable agents to acquire complex behaviors at run-time, in response to the changing demands of dynamic open environments.

Prototype Implementation

The goal of the Agentcities Initiative is to provide a platform to demonstrate the interoperation of independently authored agents that are geographically dispersed and executing within heterogeneous environments. The interoperation is accomplished through the use of open systems technologies and protocols. The protocols utilized in the Agentcities framework are those defined by the FIPA standards. I intend to utilize Agentcities as a research platform for the delivery of contextually appropriate Web Services, where the context is defined by geographic location. The proposed research overlaps well with the research goals of Agentcities as defined in; specifically listed is the desire to investigate the seamless interaction between wireless and wire line agents to dynamically compose services based on user location. The proposed research also overlaps with the objectives of three Agentcities working groups that are in the formative stages. These three groups are: Engineering Self-Organizing Applications WG, Service Description and Composition in Agentcities WG, and Ontologies and Semantics WG.

Figure 1:Proposed architecture

The architecture for the proposed research is found in Figure 1. The architecture is designed for scalability, from mobile PDA devices with wireless connectivity to resource-rich server class systems. The architecture is designed to be compatible with existing and emerging open standards; as such interoperability within open agent societies and Web Services is maximized. The major components of the architecture are:

Operationally, the mobile agent will receive its absolute GPS position from the onboard GPS receiver. The location will be consumed by an internal behavior that will communicate with the Wherehoo server. The Wherehoo server will return a set of DAML-S descriptions for services that are appropriate within a physical region. The region is defined as a circle with selectable radius, whose center point is the current location. The DAML-S description will be passed to the Home Server where they will be transformed into a Piccola program that will be executed by the Piccola execution engine. It is anticipated that the transformation will leverage the Transformation API for XML (TrAX) . The executing Piccola programs will communicate with the mobile agent via JXTA protocols and uni-directional pipes; JXTA provides the discovery services to allow the mobile agent to find the pipe endpoints on the Home Server. The result is the delivery of contextually appropriate Web Services to the mobile agent who views them as semantically described behaviors.

Alternatively, the Piccola processes could be wrapped in an agent and registered with the Home Server's Agent Management System. The Home Server's Directory Facilitator could then be used to link the mobile agent with its agent-based behaviors. However, when the behavior is no longer required, the mobile agent cannot teardown the remote agent without violating its autonomy. It is also intended that the enacted DAML-S descriptions are private internal behaviors of the mobile agent; advertising their existence via the DF defeats this intent. So, although the mobile agent and Piccola processes communicate as peers, a master-slave relationship exists between them. When the mobile agent no longer requires a behavior, it will send a teardown request to the Piccola process, which will end execution. This is perceived to be an advantage of the proposed architecture.

Impact of Research

The proposed research will enable the dynamic creation of web services from DAML-S descriptions. This technology can be used to power the semantic web by creating only those services that are needed, and only when they are needed. For example, we can envision the existence of large databases consisting of DAML-S service descriptions of services that perhaps no one offers at the time. However, when one of these services is needed it could be automatically called into being using our technology.

José M Vidal
This paper is available at
Last modified: Wed Jun 19 10:49:40 EDT 2002