Michael J. R. Shave

KRAFT: knowledge fusion from distributed databases and knowledge bases

The KRAFT project aims to investigate how a distributed architecture can support the transformation and reuse of a particular class of knowledge, namely constraints, and to fuse this knowledge so as to gain added value, by using it for constraint solving or data retrieval.

Resolving Ontological Heterogeneity in the KRAFT Project

KRAFT is an agent architecture for the integration of heterogeneous information systems. The focus in KRAFT is on the integration of knowledge in the form of constraints. In this article we describe the architecture from an ontological perspective. We start by introducing the agent architecture and illustrate its application in the telecommunication-network design. We then describe how we assess the ontological heterogeneity in the domain, which problems the integration of constraint knowledge pose, and how we construct a shared ontology. Also, we describe the mapping, functions that are used to translate information between the shared and the local ontologies. Finally, we look at the direction our research is taking hereafter.

Two aspects of the validation and verification of knowledge-based systems

Two research projects that have examined problems related to system verification and validation and the associated issue of maintenance are described. The Mekas (Methodology for Knowledge Analysis) Project has developed a method to give knowledge engineers a thorough characterization of the domain, that is, a full description of the ontology, structure, functions, and theories that underpin the domain. This provides a coherent framework within which a knowledge base can be more readily identified with key aspects of the domain, aiding verification and validation. The second project, known as MAKE (Maintenance Assistance for Knowledge Engineers), was aimed at the process of constructing and revising the knowledge base. It is argued that the primary focus of verification and validation must be on the models, and the transitions between them, rather than simply on the executable representation.<<ETX>>

Transfer of Natural Metaphors to Parallel Problem Solvin Applications

Metaphors are crucial to research as they provide alternative perspectives of seeing and understanding the world; indeed, they have been the basis of many scientific and technological creations. In this paper, we argue that progress in parallel problem solving research could be invaluably enhanced if metaphors that exhibit parallelism, in terms of their underlying structure and the ways they work, are studied. Using biological examples, we demonstrate how nature furnishes us with numerous metaphors. An attempt is made to show how an understanding of the way such metaphors work can help us manage our conceptualisation of complex problems as well as reveal new directions towards parallel problem applications.

Facilitating the development of knowledge based systems: a critical review of acquisition tools and techniques

Certain problems associated with knowledge acquisition are identified and examined in this paper. We review a variety of methodologies and tools designed to address these problems and then argue that there is a strong case for a preliminary knowledge analysis or domain phase of KBS development. This phase facilitates subsequent design, development and maintenance phases. The details of our domain characterisation are not expounded upon in this paper. The paper concludes with a suggestion of a re-examination of some of the central metaphore acquisition.

A language for simple interactive retrieval from a database system

Abstract Query interaction with a database system requires, in general, some understanding of the content and structure of the database, and knowledge of a suitable query language to encode the request for data. These factors impose barriers against access to a database on a casual or irregular basis. To overcome such restrictions we have investigated the use of a pseudo-intelligent front-end retrieval system. This system was designed to be independent of any specific database management system, although a relational database structure was considered to be the most appropriate. A prototype version of the system has been set up to run on top of Logicas relational database management system RAPPORT and the IBM relational database system SQL. As a result of this experience we have developed an adaptable language to facilitate intelligent interaction between an end user and a database management system.

An Examination of Some Metaphorical Contexts for Biologically Motivated Computing

Biologically motivated computing seeks to transfer ideas from the biosciences to computer science. In seeking to make transfers it is helpful to be able to appreciate the metaphors which people use. This is because metaphors provide the context through which analogies and similes are made and by which many scientific models are constructed. As such, it is important for any rapidly evolving domain of knowledge to have developments accounted for in these terms. This paper seeks to provide one overview of the process of modelling and shows how it can be used to account for a variety of biologically motivated computational models. Certain key ideas are identified in the subsequent analysis of biological sources, notably, systemic metaphors. Three important aspects of biological thinking are then considered in the light of computer science applications: biological organization, the cell, and models of evolution. The analysis throughout the paper is descriptive rather than formalized so that a large variety of potential applications may be considered.

Spatial Reasoning for GIS Using a Tesseral Data Representation

A technique for spatial reasoning is described which is directly compatible with the data representation techniques used by Geographical Information Systems (GIS). GIS may be thought of as specialised forms of database systems which are distinguished by their ability to handle spatial data and are widely used to aid environmental planners to make decisions and predictions. However, their application is limited by their lack of support for any spatio-temporal reasoning capability. The technique described here provides GIS with this capability. It is based on a quad tesseral addressing representation of space supported by a constraint based reasoning mechanism. The technique has been incorporated into a tesseral spatial reasoning system (SPARTA) which has been used successfully to resolve two-dimensional reasoning scenarios.

Spatial Reasoning Using the Quad Tesseral Representation

A review of the application of the quad tesseral representation tosupport spatial reasoning is presented. The principal feature of therepresentation is that it linearises multi-dimensional space, while stillproviding for the description of individual objects within that space andthe relationships that may exist between those objects (in any directionand through any number of dimensions). In addition the representation issupported by an arithmetic which allows the manipulation (translation etc.)of spatial objects. Consequently, when incorporated into a spatialreasoning system, all necessary processing can be implemented as if in onlyone dimension. This offers two significant advantages over moreconventional multi-directional approaches to spatial reasoning. Firstly,many of the concerns associated with the exponential increase in the numberor relations that need to be considered (as the number of dimensions underconsideration increases) are no longer relevant. Secondly, the computationalcost of manipulating and comparing spatial objects remains static at itsone dimensional level, regardless of the number of dimensions underconsideration.

An Ontology for Linear Spatial Reasoning

An ontology for spatial reasoning based on a tesseral representation of space is presented. The principal advantage offered is that the representation has the effect of linearising multi-dimensional space while still supporting translation through the space in any direction and through any number of dimensions. Consequently, all multi-dimensional spatial reasoning can be implemented using one dimensional (temporal) reasoning techniques. As a result, many of the concerns associated with conventional multi-dimensional spatial reasoning systems, based on more traditional representations, no longer apply.