This paper describes the development of FLAIR (Flexible Learning with an Artificial Intelligence Repository), a repository of educational material and of a highly visual computing environment for use in laboratories associated with the introductory undergraduate Artificial Intelligence (AI) course. This repository supports sharing of pedagogic material and of development tools, and cooperation in their use, while allowing diversity in content and in use at different institutions. Thus the development of the repository has stressed the production of system tools, extensible object-oriented libraries, and strong programming frameworks. Some modules currently available are on Search and Automated Reasoning. Examples of the presentation techniques used are conceptual maps, hypertext, and graphic animations of algorithms. Initial experience in the use of the repository in teaching the introductory AI course is taking place in the 1993/94 academic year.

As interactive users find conventional methods of training and documentation inadequate, designers are providing systems with online reference information, descriptions of valid input, elaboration of error messages, and explanations of a system's behavior. This paper describes some existing commercial systems that offer online assistance and more experimental approaches by the research community. The following material was originally presented at the SIGSOC conference on Easier and More Productive Use of Computing Systems. An extended version will appear in a special issue of the IEEE Transactions on Systems, Man, and Cybernetics (Volume SMC-12, March/April, 1982), and is reprinted here with the permission of the IEEE.Online user assistance is now offered on commercial systems and is the subject of investigation in experimental settings. It is difficult to compare the advantages and limitations of different approaches because they vary along many dimensions and because there is no commonly accepted terminology. A grouping of these dimensions into major categories is a necessary first step towards more empirical evaluations. The major software-related features of online assistance appear to fall into four categories:• access method -- the way users can construct or enter requests for assistance;• data structure -- the manner in which different portions of assistance information are related to each other;• software architecture -- how assistance requests and their responses are communicated among a user, an operating system, application programs, and the assistance database; and• contextual knowledge -- how much information is retained about the assistance environment, including the user, the application, and the tasks being performed.

A cost is defined for demand paging algorithms with respect to a formal stochastic model of program behavior. This cost is shown to exist under rather general assumptions, and a computational procedure is given which makes it possible to determine the optimal cost and optimal policy for moderate size programs, when the formal model is known and not time dependent. In this latter case it is shown that these computational procedures may be extended to larger programs to obtain arbitrarily close approximations to their optimal policies. In previous models either unwarranted information is assumed beyond the formal model, or the complete stochastic nature of the model is not taken into account.