A realistic model of decision making must account for a multitude of goals and limited resources.... more A realistic model of decision making must account for a multitude of goals and limited resources. Many problems will not admit to perfect solutions. Practical decisions often involve trade-offs. We present a goal-based model of decision making in which the relative priorities of goals drive the decision process. This model serves as the basis for the VOTE program which simulates Congressional roll call voting decisions on thousands of votes, based on the relative priorities of goals and constituent relationships, and the need to provide an explanation for the decision. Alternative models of decision making have used prescriptive decision analysis, which focuses on estimates of payoffs and uncertainty of outcomes. The assumptions of decision analysis are inappropriate in many domains for which our goal-based, descriptive model may apply. In developing the role of goals and priorities in decision making, we explore a number of significant dimensions of goals, resulting in a broad range of types of goals that share basic, canonical properties related to decision making. We define the importance of a goal to be proportional to the resources that an agent may choose to allocate in service to that goal. Resources are viewed quite broadly to include not only physical resources, but also cognitive resources, such as attention and memory. We extend the model of goals to include interpersonal relationships. An agent adopts the goal agenda of another agent with a priority proportional to the importance of the relationship. This extension permits the processing of adopted goals in a manner uniform with individual goals. The VOTE program does not rely on a single method for ranking alternatives, but rather implements a collection of strategies that relate decisions to explanations. VOTE's knowledge base is large and modular, built on top of an object-oriented database system. VOTE provides a natural language generation facility that makes it easier to interpret the results. VOTE can produce explanations and interpretations in both English and French. The VOTE program is a robust demonstration of the feasibility of the decision making model presented here. The model might also be applied to other resource allocation domains, as well as other cognitive tasks including planning, learning, and understanding.
Contents: Goals and Decisions. Overview of the VOTE Program. Goals. Resources. Interpersonal Rela... more Contents: Goals and Decisions. Overview of the VOTE Program. Goals. Resources. Interpersonal Relations. The VOTE Program. Decision Strategies. Related Work. Future Work. Appendices: Natural Language Generation. Inside VOTE. Issues. Bills. Groups. Members.
Abstract: Our work at the Yale Artificial Intelligence Project has provided key scientific insigh... more Abstract: Our work at the Yale Artificial Intelligence Project has provided key scientific insights into human cognitive behavior in general and learning in particular. These scientific developments can be applied to education. At the heart of thinking is the need to explain ...
A realistic model of decision making must account for a multitude of goals and limited resources.... more A realistic model of decision making must account for a multitude of goals and limited resources. Many problems will not admit to perfect solutions. Practical decisions often involve trade-offs. We present a goal-based model of decision making in which the relative priorities of goals drive the decision process. This model serves as the basis for the VOTE program which simulates Congressional roll call voting decisions on thousands of votes, based on the relative priorities of goals and constituent relationships, and the need to provide an explanation for the decision. Alternative models of decision making have used prescriptive decision analysis, which focuses on estimates of payoffs and uncertainty of outcomes. The assumptions of decision analysis are inappropriate in many domains for which our goal-based, descriptive model may apply. In developing the role of goals and priorities in decision making, we explore a number of significant dimensions of goals, resulting in a broad range of types of goals that share basic, canonical properties related to decision making. We define the importance of a goal to be proportional to the resources that an agent may choose to allocate in service to that goal. Resources are viewed quite broadly to include not only physical resources, but also cognitive resources, such as attention and memory. We extend the model of goals to include interpersonal relationships. An agent adopts the goal agenda of another agent with a priority proportional to the importance of the relationship. This extension permits the processing of adopted goals in a manner uniform with individual goals. The VOTE program does not rely on a single method for ranking alternatives, but rather implements a collection of strategies that relate decisions to explanations. VOTE's knowledge base is large and modular, built on top of an object-oriented database system. VOTE provides a natural language generation facility that makes it easier to interpret the results. VOTE can produce explanations and interpretations in both English and French. The VOTE program is a robust demonstration of the feasibility of the decision making model presented here. The model might also be applied to other resource allocation domains, as well as other cognitive tasks including planning, learning, and understanding.
Contents: Goals and Decisions. Overview of the VOTE Program. Goals. Resources. Interpersonal Rela... more Contents: Goals and Decisions. Overview of the VOTE Program. Goals. Resources. Interpersonal Relations. The VOTE Program. Decision Strategies. Related Work. Future Work. Appendices: Natural Language Generation. Inside VOTE. Issues. Bills. Groups. Members.
Abstract: Our work at the Yale Artificial Intelligence Project has provided key scientific insigh... more Abstract: Our work at the Yale Artificial Intelligence Project has provided key scientific insights into human cognitive behavior in general and learning in particular. These scientific developments can be applied to education. At the heart of thinking is the need to explain ...
Uploads