|
Much of the research on intelligent agents has focused on technology issues and specific subsystems of intelligent enterprises. Zhu, Perietula, and Hsu (1997) explored building a link between information technology and organizational learning, focusing on creating business processes that can learn in a manufacturing environment. Peng et al. (1998) created a multi-agent framework that supports the integration of various types of expertise in an intelligent enterprise. Agents with domain expertise have been developed to cooperate and interact with one another as well as with human managers to reach timely decisions. Jain, Aparicio, and Singh (1999) introduced intelligent agents for streamlining processes in virtual organizations. Intelligent agents are particularly effective in providing a reliable and flexible foundation for virtual organizations through facilitating knowledge query and communications (O'Leary, Kuokka, & Plant, 1997). Sikora and Shaw (1998) proposed a multi-agent framework for integrating various traditional information systems.
The traditional view of organizational infrastructure is that it is a piece of the puzzle. Although this view highlights the importance that the infrastructure must fit in with other components of the system, it fails to reveal that infrastructure, by definition, is the foundation upon which an organization is built. The intelligence infrastructure needs to be an integrated part of all organizational functions. Figure 2 shows a simple framework of agent-supported intelligence infrastructure. Central to the intelligence infrastructure is the multi-agents system that is closely integrated with traditional information, knowledge, and business process systems.
Figure 2: Agent-Supported Intelligence Infrastructure Framework.
In traditional systems, many of the functions and processes may not be autonomous. Furthermore, interoperability may be limited due to different platforms and standards used by various subsystems. Recent developments in standards based networking and data communications, such as transactions based on SOAP (Simple Object Access Protocol) and XML (Extensible Markup Language) Web services, promise drastic improvements in machine-to-machine communications. These developments benefit standards-based intelligent agents' communications and the integration of intelligent agent systems with traditional systems.
Jack Ring, head of consulting firm Innovation Management, proposed eleven components that comprise a business enterprise. Adopting this organizational framework, we enlist various specialized intelligent agents that can be deployed in each component of the system in Table 2.
Enterprise Component | Explanation | Intelligent Agents |
---|---|---|
Mission | Statement of purpose | Environment scanning Executive support |
Vision | Perceived future status | Environment scanning Executive support Business intelligence Strategic planning |
Resources | Non human assets | Planning and scheduling Resource management Monitoring Diagnosis |
Policies | Explicit rules and procedures | Info retrieving and filtering Planning and scheduling Resource management Monitoring |
Processes | Defined activities and sequences | Process control Procurement Planning and scheduling Knowledge codification Knowledge distribution |
Information | Meaningful and accessible data | Information collection Information retrieving and filtering Intranet and Web crawler Directory and category Ontology creation and support |
Systems | Sets of goal oriented and coordinated components | Enterprise resource planning Supply chain management Data mining Customer relationship management Network security E-commerce E-business |
Culture | Implicit beliefs, values, and rituals | Training and tutoring Community support Virtual mentors |
People | "DNA of the business" The ultimate decision makers | Interface Communication assistant Personal assistant Collaborative work Decision support Project management Affective Training and tutoring Knowledge acquisition |
Products | What the customers want | Business intelligence R&D subsystem Computer aided design Visualization Simulation |
Services | What the customers need | Business intelligence Contract negotiation Contract management Outsourcing |
Many of the intelligent agents listed in the table, such as Planning and scheduling, Information filtering, and Interface agents, have already been developed. A good summary of intelligent agents applications in various business functional areas can found in Intelligent Software Agents (Murch & Johnson 1999). Others types of agents, such as Environment scanning, Knowledge acquisition, and Ontology creation, are proposed to support various organizational functions that are essential to intelligent enterprises. Table 2 serves as a visual map that links the role of intelligent agents to organizational components.
Although the multi-agent system is integrated with traditional systems in an intelligence infrastructure, a three-tier model that treats the agent-based system as an intermediary between users and services can help to highlight the role of intelligent agents. Intelligent agents work autonomously for their masters (and are controlled by their masters, if necessary). Various agents play different roles to provide a wide range of intermediary services such as controlling and monitoring system operations, retrieving and filtering information, and negotiating and interacting with other users and/or other intelligent agents.
The end user does not need to know the design and the inner work of the intelligence infrastructure or the information and services available. Through interface agents, the user can query the system, issue commands, and request services. The behavior of interface agents can be customized to meet the user's preferences. Intelligent interface agents can also learn and adapt to the styles of their masters through observing the user's actions. In a nutshell, the user interacts with a personalized interface agent that knows how to satisfy his/her needs.
Interface agents, in turn, interact and collaborate with other intelligent agents to accomplish user-requested tasks. Task-specific agents work in the distributed system either individually or collectively to obtain the information or services. Although today's information and knowledge systems are mostly networked, cross-system communication and sharing are typically limited to simple information retrieval and message exchanges. Integration of multi-agent systems with existing information infrastructure makes it possible to develop distributed intelligence across the entire organization.
The distributed intelligence can be extended to the extranet and the Internet, as Internet-based network standards (TCP/IP, XML, SOAP, etc.) gain wider acceptance in the corporate world. For example, business intelligence agents can crawl the Web to gather information. Because of built-in intelligence, agents are capable of searching information more effectively than search engines. The intelligence infrastructure can be extended to mobile workers through lightweight mobile computers and mobile intelligent agents. Mobile intelligent agents can migrate though computer networks in order to satisfy requests made by the user.
Figure 3: Three Tier Model of Intelligence Infrastructure.
|