Robert St. Amant

AAAI Conference 2006 Conference Paper

AI Support for Building Cognitive Models

• Robert St. Amant

Cognitive modeling techniques provide a way of evaluating user interface designs, based on what is known about human cognitive strengths and limitations. Cognitive modelers face a tradeoff, however: more detailed models require disproportionately more time and effort to develop than coarser models. In this paper we describe a system, G2A, that automatically produces translations from abstract GOMS models into more detailed ACT-R models. G2A demonstrates how even simple AI techniques can facilitate the construction of cognitive models and suggests new directions for improving modeling tools.

AAAI Conference 2005 Conference Paper

Tool Use for Autonomous Agents

• Robert St. Amant

The intelligent use of tools is a general and important human competence that AI research has not yet examined in depth. Other fields have studied the topic, however, with results we can compile into a broad characterization of habile (tool-using) agents. In this paper we give an overview of research on the use of physical tools, using this information to motivate the development of artificial habile agents. Specifically, we describe how research goals and methods in animal cognition overlap with those in artificial intelligence. We argue that analysis of activities of tool-using agents offers an informative way to evaluate intelligence.

AAMAS Conference 2003 Conference Paper

Social Navigation: Modeling, Simulation, and Experimentation

• Mark O. Riedl
• Robert St. Amant

AAAI Conference 1997 Conference Paper

Navigation and Planning in a Mixed-Initiative User Interface

• Robert St. Amant

Mixed-initiative planning is one approach to building an intelligent decision-making environment. A mixedinitiative system shares decision-making responsibility with the user such that it acts sometimes as a tool, to be directly applied to a specific task, and other times as an autonomous problem-solver. In the best case, the user can delegate the details of a task to the automated system without giving up the ability to guide and review the decision-making process. We have developed a simple mixed-initiative planner that incorporates a view of problem-solving as navigation. We have explored this notion in two different applications: exploratory statistical analysis and layout design for user interface dialogs. This paper discusses navigation issues in the context of these two systems, the potential benefits of the approach, and some implications for user interface design.

ICAPS Conference 1996 Conference Paper

A Planner for Exploratory Data Analysis

• Robert St. Amant
• Paul R. Cohen

Statistical exploratory data analysis (EDA) poses a difficult search problem. The EDA process lends itself however to a planning formulation. We have built a system, called AIDE, to help users explore data. AIDE relies on partial hierarchical planning, a form of planning appropriate for tasks in complex. uncertain environments. Our description of the EDA task and the AIDE system provides a case study of the applicat ion of plamdng to a novel domain.