Abstract not found

Levesque and Brachman argue that in order to provide timely and correct responses in the most critical applications, general purpose knowledge representation systems should restrict their languages by omitting constructs which require non-polynomial worst-case response times for sound and complete classification. They also separate terminological and assertional knowledge, and restrict classification to purely terminological information. We demonstrate that restricting the terminological language and classifier in these ways limits these "general-purpose" facilities so severely that they are no longer generally applicable. We argue that logical soundness, completeness, and worst-case complexity are inadequate measures for evaluating the utility of representation services, and that this evaluation should employ the broader notions of utility and rationality found in decision theory. We suggest that general purpose representation services should provide fully expressive languages, classi...

W’e present a formal s \ ntax and semantics for SNePS considered as the (modeled) mind of a cogn:ti\e agent. The semantics is based on a Meinongian theory of ’ the intensional objects of ’ thought that is appropriate for 41 considered as “computational philosophy ” or “computational psychology”. 1. INTRODUCTION. W’e present a formal syntax and semantics for the SNePS Semantic Network P recessing System (Shapiro 1979), based on a \leinongian theory of the intensional objects of thought (Rapaport 198Sa). Such a theory avoids possible worlds and is appropriate

The representation formalism as well as the representation language is of great importance for the success of machine learning. The representation formalism should be expressive, efficient, useful, and applicable. First-order logic needs to be restricted in order to be efficient for inductive and deductive reasoning. In the field of knowledge representation term subsumption formalisms have been developed which are efficient and expressive. In this paper, a learning algorithm, KLUSTER, is described which represents concept definitions in this formalism. KLUSTER enhances the representation language if this is necessary for the discrimination of concepts. Hence, KLUSTER is a constructive induction program. KLUSTER builds the most specific generalization and a most general discrimination in polynomial time. It embeds these concept learning problems into the overall task of learning a hierarchy of concepts.

There is a subtle risk of ambiguity in the choice between concepts and roles forced by current KL-ONE-like languages, since many roles may be concepts as well. In this paper we explore the ontological foundations of the role/concept relationship, and analyze its implications on the practice of knowledge engineering. We criticize the current interpretation of KL-ONE roles as arbitrary relations, which vanishes their original meaning and makes them identical to slots. We suggest to call attributes those concepts which actually act as conceptual components, and propose a formal semantics which binds these concepts to their corresponding relations.

Based on the research done in the last decade, attempts have been made to propose description logics as unifying formalisms for the various class-based representation languages used in different areas. These attempts have made apparent that sound, complete, and decidable description logics still suffer from several limitations, regarding modeling classes of aggregate objects, expressing general inclusion axioms, and the ability of navigating links between classes. In this paper we make description logics accomplish the necessary leap in order to become suitable for the new challenging applications they are faced with. In particular, we propose a powerful description logic overcoming the above limitations and we show that its reasoning tasks are decidable in worst case exponential time. 1

In the past 20 years, AI researchers in knowledge representation (KR) have implemented over 50 frame knowledge representation systems (FRSs). KR researchers have explored a large space of alternative FRS designs. This paper surveys the FRS design space in search of design principles for FRSs. The FRS design space is defined by the set of alternative features and capabilities --- such as the representational constructs --- that an FRS designer might choose to include in a particular FRS, as well as the alternative implementations that might exist for a particular feature. The paper surveys the architectural variations explored by different system designers for the frame, the slot, the knowledge base, for accessoriented programming, and for object-oriented programming. We find that few design principles exist to guide an FRS designer as to how particular design decisions will affect qualities of the resulting FRS, such as its worst-case and average-case theoretical complexity, its actual...

Concept mapping has a history of use in many disciplines as a formal or semi-formal diagramming technique. Concept maps have an abstract structure as typed hypergraphs, and computer support for concept mapping can associate visual attributes with node types to provide an attractive and consistent appearance. Computer support can also provide interactive interfaces allowing arbitrary actions to be associated with nodes such as hypermedia links to other maps and documents. This article describes a general concept mapping system that is open architecture for integration with other systems, scriptable to support arbitrary interactions and computations, and cutomizable to emulate many styles of map. The system supports collaborative development of concept maps across local area and wide area networks, and integrates with World-Wide Web in both client helper and server gateway roles. A number of applications are illustrated ranging through education, artificial intelligence, active documents...

We present a knowledge-based approachtointelligent information integration based on a re-classi#cation of information entities in a new context. We identify the needs of approaches performing semantic translation, weintroduce a uniform description of contexts and discuss a naive classi#cation approach. We argue that this approach makes unrealistic assumptions about the absence of uncertainty. Toovercome this problem we discuss several approximate classi#cation approaches and their use for information integration. Therebywe address symbolic as well as numeric approaches for uncertainty handling. We sumarize with a description of an actual application area and a discussion of open research topics.

classic is a recently developed knowledge representation (KR) system, based on a view of frames as structured descriptions, with several important inferable relationships, including description classification. While much about classic is novel and important in its own right, it is especially interesting to consider the system in light of its unusual (for Artificial Intelligence) intellectual history: it is the result of over a decade of research and evolution in representation systems that trace their origins back to work on kl-one, arguably one of the most long-lived and influential approaches to KR in the history of AI. We outline some of the novel contributions of classic, but pay special attention to its roots, illustrating the maturation of some of the original features of kl-one, and the decline and fall of others. A number of key ideas are analyzed---including the interpretation of frames as descriptions, the classification inference, and the role of a knowledge representation s...