base that pertains to each individual and training domain. We employ two types of logobots: student logobot and instructor logobot. The student logobot acts as the student's personalized interface to the remote training. It uses an acquired knowledge base of preferences and reference material and maintains each student's bookmarks. Salient features enabled by this agent include • On-demand or automated update of training material. • An intuitive interface to the training material, which includes the ability to search the student's bookmarks by specifying domain, URL, title, or description. • Personalization by selection of training domains and other preferences. • Ability to change the default preferences assigned by the instructor. These capabilities provide flexibility for students to identify their training needs. This not only motivates students but also helps them explore the paradigm of learning to learn and just-in-time learning. The instructor logobot acts as both an interface agent and an information agent. It helps instructors easily acquire training information from various sources and automatically disseminate it appropriately to the students. It can search the Internet and helps instructors filter relevant information and classify it into different domains. It also assigns default training preferences for different classes of students. Multiple instructor logobots maintain a centralized knowledge base for various training domains. Each instructor logobot T he U.S. military has several training facilities geared toward military personnel who are geographically distributed. Like other military units, the U.S. Army offers mail-based correspondence courses from battle laboratories such as Combined Arms Support Command (CASCOM) in Fort Lee, Va. Although there is no core curriculum, students enroll in different courses based on their military occupational specialty (MOS). The project described in this article is indicative of the steps the Army is taking to migrate from paper-based to electronic training by way of the Internet. This project provides instructors with tools to gather and categorize Internet uniform resource locators (URLs), and lets students choose categories of URLs that can be automatically stored in their bookmark (favorites) collection. The instructor can exercise control over training materials accessed by students and, as a secondary benefit, can restrict unnecessary " net surfing. " Because one of the aims was to provide interdisciplinary training to students, the subject categories for organizing the URLs, which we labeled domains, cut across the students' MOS categorization. They can subscribe to any domain they wish, but can only access the URLs provided by the …
[1]
Mary Shaw,et al.
Software architecture - perspectives on an emerging discipline
,
1996
.
[2]
Nicholas R. Jennings,et al.
Intelligent agents: theory and practice
,
1995,
The Knowledge Engineering Review.
[3]
Michael R. Genesereth,et al.
Software agents
,
1994,
CACM.
[4]
Arthur C. Graesser,et al.
Is it an Agent, or Just a Program?: A Taxonomy for Autonomous Agents
,
1996,
ATAL.
[5]
Pattie Maes,et al.
Agents that reduce work and information overload
,
1994,
CACM.
[6]
Sarosh N. Talukdar,et al.
Multiagent organizations for real-time operations
,
1992,
Proc. IEEE.