Artificial Intelligence Review 14: 121–141, 2000. Stephen J. Hegner et al. (eds.), Intelligent help systems for UNIX. © 2000 Kluwer Academic Publishers. Printed in the Netherlands. 121 USCSH: An Active Intelligent Assistance System MANTON MATTHEWS * , WALTER PHARR 1 , GAUTAM BISWAS 2 and HARISH NEELAKANDAN Department of Computer Science, University of South Carolina, Columbia, SC 29208, USA, matthews@cs.sc.edu; 1 Current address: College of Charleston, Charleston, SC 29424, USA; 2 Current address: Vanderbilt University, Nashville, TN37235, USA ( * author for correspondence) Abstract. This paper describes the knowledge sources and methods of knowledge acquisition for USCSH (University of South Carolina SHell). USCSH is an active intelligent assistance system for Unix. 1 The system operates in two modes, the active mode and the intelligent mode. In the active mode USCSH monitors the user’s interactions with the system, and at appropriate times makes suggestions on how the user may better utilize the system to perform tasks. In the intelligent mode the system accepts questions in natural language and responds to them, taking into consideration the ability of the user and the context of the question. Keywords: assistance systems, shell, user-modelling 1. Introduction New software systems are constantly increasing in functionality, and the continued increase in hardware capability-to-cost ratio will serve to make the delivery of more sophisticated software possible. This makes an assist- ance system an essential component of these complex systems. Traditional approaches to assistance generally fall into one of several categories: com- mand indexed systems (e.g., Unix man), hierarchical systems (e.g., VMS help), and tutorials (e.g., Unix learn). Regardless of the category these traditional approaches to assistance gen- erally suffer from several deficiences. Responses are too verbose and are not adapted to the user’s level of knowledge of the system. On-line manuals must explain everything, and novices find them confusing, while more experienced users find it quite annoying to have to browse through a lot of irrelevant material to find the information they desire. Moreover, the systems only respond when the users realize that they need to know something, and only after they figure out the right keyword to use in the query (which places the burden of translating a request for specific help into command names on the user), or by browsing through menus (which