Stuart C. Shapiro

A model for belief revision

Abstract It is generally recognized that the possibility of detecting contradictions and identifying their sources is an important feature of an intelligent system. Systems that are able to detect contradictions, identify their causes, or readjust their knowledge bases to remove the contradiction, called Belief Revision Systems, Truth Maintenance Systems, or Reason Maintenance Systems, have been studied by several researchers in Artificial Intelligence (AI). In this paper, we present a logic suitable for supporting belief revision systems, discuss the properties that a belief revision system based on this logic will exhibit, and present a particular implementation of our model of a belief revision system. The system we present differs from most of the systems developed so far in three respects: First, it is based on a logic that was developed to support belief revision systems. Second, it uses the rules of inference of the logic to automatically compute the dependencies among propositions rather than having to force the user to do this, as in many existing systems. Third, it was the first belief revision system whose implementation relies on the manipulation of sets of assumptions, not justifications.

THE SNePS SEMANTIC NETWORK PROCESSING SYSTEM

Publisher Summary This chapter describes the SNePS semantic network processing system, which is a direct descendent of MENTAL. SNePS is currently implemented in ALISP and runs interactively on the CDC CYBER 173 at the State University of New York at Buffalo. There are three levels of SNePS: (1) the abstract graph level, (2) the pictorial level, and (3) the linear symbolic level. For the latter, the SNePS User Language, SNePSUL, is used. Semantic networks have been used as representations of knowledge since the mid-1960s. There are four levels at which semantic networks can be discussed: (1) an abstract graph level, (2) a two-dimensional pictorial level, (3) a one-dimensional symbolic level, and (4) a computer implementation level. These levels, though related, are independent in the sense that two semantic networks can differ on one or more levels and be the same on the other levels.

Intensional concepts in propositional semantic networks

An integrated statement is made concerning the semantic status of nodes in a propositional semantic network, claiming that such nodes represent only intensions. Within the network, the only reference to extensionality is via a mechanism to assert that two intensions have the same extension in some world. This framework is employed in three application problems to illustrate the nature of its solutions. The formalism used here utilizes only assertional information and no structural, or definitional, information. This restriction corresponds to many of the psychologically motivated network models. Some of the psychological implications of network processes called node merging and node splitting ore discussed. Additionally, it is pointed out that both our networks and the psychologically based networks are prone to memory confusions about knowing unless augmented by domain-specific inference processes, or by structural information.

The SNePS Family

Abstract SNePS, the Semantic Network Processing System, is an intensional propositional semantic network that has been designed to be the mind of a computational cognitive agent. In this article, the main features of SNePS are sketched, its antecedents are discussed, and some example current uses are described.

Breast cancer screening in 21 countries: delivery of services, notification of results and outcomes ascertainment.

Following clinical trial evidence of mammography screenings efficacy and effectiveness, data are needed from organized population-based programmes to determine whether screening in these programmes results in breast cancer mortality reductions comparable to those demonstrated in controlled settings. The International Breast Cancer Screening Network (IBSN) conducted two international programme assessments: in 1990 among nine countries and in 1995 among 22 countries, obtaining information on the organization and process for screening within breast cancer screening programmes. This manuscript describes procedures for recruitment, service delivery, interpretation and communication of results, case ascertainment, and quality assurance. Practices in more established programmes are compared with pilot programmes. Each IBSN country defined a unique programme of population-based breast cancer screening. Some programmes were sub-national rather than national in scope, while others were in pilot stages of development. Screening took place in dedicated centres in established programmes and in both dedicated and general radiology centres in pilot programmes. Although most countries used personal invitation systems to recruit women to screening, other recruitment mechanisms were used. Most countries used two-view mammography in their screening programmes. About half had implemented independent double reading of mammograms, considering it a key component of high-quality mammography screening. In conclusion, diversity exists in the organization and delivery of screening mammography internationally. Quality assurance activities are a priority and are being evaluated in the IBSN.

Anchoring in a grounded layered architecture with integrated reasoning

The GLAIR grounded layered architecture with integrated reasoning for cognitive robots and intelligent autonomous agents has been used in a series of projects in which Cassie, the SNePS cognitive agent, has been incorporated into hardware- or software-simulated cognitive robots. In this paper, we present an informal, but coherent, overview of the GLAIR approach to anchoring the abstract symbolic terms that denote an agent’s mental entities in the lower-level structures used by the embodied agent to operate in the real (or simulated) world. We discuss anchoring in the domains of: perceivable entities and properties, actions, time, and language.

Belief spaces as sets of propositions

Abstract It is common in the knowledge representation literature for a belief space to be considered to be a set of sentences. Some implications of this stance are examined, and an alternative view, that belief spaces are sets of propositions is developed, and found to be an improvement. This latter view requires that propositions be accepted as entities in the domain of discourse of languages of thought, which, it is argued, accords with commonsense usage. In exchange, the semantics of nested belief expressions is simplified, and certain problems caused by the sentential view are avoided.

Mapping the Landscape of Human-Level Artificial General Intelligence

We present the broad outlines of a roadmap toward human-level artificial general intelligence (henceforth, AGI). We begin by discussing AGI in general, adopting a pragmatic goal for its attainment and a necessary foundation of characteristics and requirements. An initial capability landscape will be presented, drawing on major themes from developmental psychology and illuminated by mathematical, physiological and information processing perspectives. The challenge of identifying appropriate tasks and environments for measuring AGI will be addressed, and seven scenarios will be presented as milestones suggesting a roadmap across the AGI landscape along with directions for future research and collaboration.

The CASSIE Projects: An Approach to Natural Language Competence

The CASSIE projects are united not only by shared computer systems and techniques, but by the philosophy that one powerful way to create intelligent systems is to create systems that can be instructed via natural language (and NL extended with graphics and gestures) what to believe, how to reason, and how to behave. All the versions of CASSIE analyze inputs with respect to stored beliefs that are modified by the inputs, and generate output based on current beliefs.

CABLES, PATHS, AND “SUBCONSCIOUS” REASONING IN PROPOSITIONAL SEMANTIC NETWORKS

Publisher Summary The definition of SNePS molecular nodes as cablesets captures the notion that a new arc (wire) cannot be added as emanating from an already existing node. This would amount to changing the denotation of the node. Instead, a new wire joined to an old node makes a new node that is related to the old one by the reduction relation. Similarly, a node without one or more of its wires is a new node that is a reduction of the old one. The propositions denoted by a node and a reduction of it are related by reduction inference, which is one kind of subconscious inference supported by SNePS. Path-based inference is another kind of subconscious inference that justifies belief in a proposition when a reduction is already believed and the extra wires are inferred from path-based inference rules and paths in the network. The set of propositions subconsciously believed by the SNePS agent is the set denoted by the set of nodes that could be gotten by path-based closure of asserted nodes followed by reduction. These nodes are virtually or implicitly in the net, and need be made explicit only when there is a specific reason.