In this paper we develop a representation for the temporal structure inherent in human actions and demonstrate an effective method for using that representation to detect the occurrence of actions. The temporal structure of the action, sub-actions, events, and sensor information is described using a constraint network based on Allen's interval algebra. We map these networks onto a simpler, 3-valued domain (past,now,fut) network --- a PNF-network --- to allow fast detection of actions and sub-actions. The occurrence of an action is computed by considering the minimal domain of its PNF-network, under constraints imposed by the current state of the sensors and the previous states of the network. We illustrate the approach with examples, showing that a major advantage of PNF propagation is the detection and removal of situations inconsistent with the temporal structure of the action. We also examine a method to increase the robustness of PNF-propagation in the case of faulty sensors. 1 In...

Combinatorial explosion of inferences has always been a central problem in artificial intelligence. Although the inferences that can be drawn from a reasoner's knowledge and from available inputs is very large (potentially infinite), the inferential resources available to any reasoning system are limited. With limited inferential capacity and very many potential inferences, reasoners must somehow control the process of inference. Not all inferences are equally useful to a given reasoning system. Any reasoning system that has goals (or any form of a utility function) and acts based on its beliefs indirectly assigns utility to its beliefs. Given limits on the process of inference, and variation in the utility of inferences, it is clear that a reasoner ought to draw the inferences that will be most valuable to it. This paper presents an approach to this problem that makes the utility of a (potential) belief an explicit part of the inference process. The method is to generate exp...

Abduction is often viewed as inference to the "best" explanation. However, the evaluation of the goodness of candidate hypotheses remains an open problem. Most artificial intelligence research addressing this problem has concentrated on syntactic criteria, applied uniformly regardless of the explainer's intended use for the explanation. We demonstrate that syntactic approaches are insufficient to capture important differences in explanations, and propose instead that choice of the "best" explanation should be based on explanations' utility for the explainer 's purpose. We describe two classes of goals motivating explanation: knowledge goals reflecting internal desires for information, and goals to accomplish tasks in the external world. We describe how these goals impose requirements on explanations, and discuss how we apply those requirements to evaluate hypotheses in two computer story understanding systems. In order to learn from experience, a reasoner must be able to explain what...

A fundamental issue in case-based reasoning is similarity assessment: determining similarities and differences between new and retrieved cases. Many methods have been developed for comparing input case descriptions to the cases already in memory. However, the success of such methods depends on the input case description being sufficiently complete to reflect the important features of the new situation, which is not assured. In case-based explanation of anomalous events during story understanding, the anomaly arises because the current situation is incompletely understood; consequently, similarity assessment based on matches between known current features and old cases is likely to fail because of gaps in the current case's description. Our solution to the problem of gaps in a new case's description is an approach that we call constructive similarity assessment. Constructive similarity assessment treats similarity assessment not as a simple comparison between fixed new and old cases, b...

explanation schemas for why people do things. These are standard high-level explanations for actions, such as "Actor does action because the outcome of action satisfies a goal of the actor." 2. Explanatory cases. These are specific explanations for particular situations, such as "Shiite Moslem religious fanatic goes on suicide bombing mission." For example, an explanation of type 1 for story S-2 might be "Because she wanted to destroy the Israeli base more than she wanted to stay alive." An explanation of type 2 would be simply "Because she was a religious fanatic." The internal causal structure of the latter explanation could then be elaborated to provide a detailed motivational analysis in terms of explanations of the first type if necessary. Both types of explanatory knowledge are represented using volitional XPs with the internal structure discussed earlier. Volitional XPs relate the actions in which the characters in a story are involved to the outcomes that those actions had for ...

This paper presents a theory of motivational analysis, the construction of volitional explanations to describe the planning behavior of agents. We discuss both the content of such explanations, as well as the process by which an understander builds the explanations. Explanations are constructed from decision models, which describe the planning process that an agent goes through when considering whether to perform an action. Decision models are represented as explanation patterns, which are standard patterns of causality based on previous experiences of the understander. We discuss the nature of explanation patterns, their use in representing decision models, and the process by which they are retrieved, used and evaluated. 1 Issues in explanation In order to learn from experience, a reasoner must be able to explain what it does not understand. When a novel or poorly understood situation is processed, it is interpreted in terms of knowledge structures already in memory. As long as thes...

The Letter Spirit project is an attempt to model central aspects of human high-level perception and creativity on a computer, focusing on the creative act of artistic letter-design. The aim is to model the process of rendering the 26 lowercase letters of the roman alphabet in many different, internally coherent styles. Two important and orthogonal aspects of letterforms are basic to the project: the categorical sameness possessed by instances of a single letter in various styles (e.g., the letter `a' in Baskerville, Palatino, and Helvetica) and the stylistic sameness possessed by instances of various letters in a single style (e.g., the letters `a', `b', and `c' in Baskerville). Starting with one or more seed letters representing the beginnings of a style, the program will attempt to create the rest of the alphabet in such a way that all 26 letters share the same style, or spirit. Letters in the domain are formed exclusively from straight segments on a grid in order to make decisions ...

Intelligent information retrieval (IIR) requires inference. The number of inferences that can be drawn by even a simple reasoner is very large, and the inferential resources available to any practical computer system are limited. This problem is one long faced by AI researchers. In this paper, we present a method used by two recent machine learning programs for control of inference that is relevant to the design of IIR systems. The key feature of the approach is the use of explicit representations of desired knowledge, which we call knowledge goals. Our theory addresses the representation of knowledge goals, methods for generating and transforming these goals, and heuristics for selecting among potential inferences in order to feasibly satisfy such goals. In this view, IIR becomes a kind of planning: decisions about what to infer, how to infer and when to infer are based on representations of desired knowledge, as well as internal representations of the system's inferential abilities ...

In addition to learning new knowledge, a system must be able to learn when the knowledge is likely to be applicable. An index is a piece of information which, when identified in a given situation, triggers the relevant piece of knowledge (or schema) in the system's memory. We discuss the issue of how indices may be learned automatically in the context of a story understanding task, and present a program that can learn new indices for existing explanatory schemas. We discuss two methods using which the system can identify the relevant schema even if the input does not directly match an existing index, and learn a new index to allow it to retrieve this schema more efficiently in the future. 1 Introduction: The index learning problem Knowledge plays an important role in AI systems, both systems that understand natural language stories and those that solve problems. However, we cannot expect systems to have all the necessary knowledge a priori. We need systems that can learn through exper...

In this paper we present a specialization of Allen interval networks that permits the rapid determination as to whether a given interval must be occurring at the current point in time. The Allen-closure of the interval network is projected into a 3-valued (past, now, fut) constraint network called a PNF-network. We show that the minimal domain of a PNF-network can be approximately computed in linear time by using arcconsistency. This computation is the key factor in the PNF propagation method of determining, for each instant of time and given information from perceptual sensors, the PNF-state of each interval, that is, happening (now), already happened (past), or has not happened (fut). We show how the computation of PNF-states can support both action recognition and the control of real-time interactive environments in which the actions are described by Allen interval networks.