Abstract

This paper describes the development of the Injection Moulding Process Expert System (IMPRESS). The IMPRESS system diagnoses faults in injection moulding machinery which lead to dirt or other contamination appearing in the plastic mouldings which are produced. This KBS has recently been put into use at Plastic Engineers (Scotland) Ltd, and is proving useful both as an expert assistant when technical help is otherwise unavailable, and as a training aid.

The IMPRESS system was built by a member of Plastic Engineers' staff with assistance from a KBS consultant. It was decided that the project would be based around a KBS methodology; a ‘pragmatic’ version of the KADS methodology was chosen. The methodology was used not only to formalise and guide the development of the KBS itself, but also to act as a framework for dividing the work between the two members of the project team. By gaining an understanding of the methodology, the staff member from Plastic Engineers was able to understand the knowledge analysis and KBS design documents produced by the consultant, and to use these documents to implement part of the KBS, both during the development of the system and when system maintenance was required.

The use of a methodology for this project on this project had both benefits and weaknesses, which are discussed at the end of the paper.

Introduction

In January 1992, Plastic Engineers (Scotland) Ltd obtained funding from Scottish Enterprise to help them in the development of a knowledge based system (KBS) for fault diagnosis.

ABSTRACT

This paper describes the structure and components of a case-based scheduler named CBS-1 which is being created to demonstrate the feasibility and utility of case-based reasoning (CBR) for dynamic job-shop scheduling problems. The paper describes the characteristics of a specific real-world scheduling task used in the work on CBS-1, identifies major problems to consider, and gives arguments for and against the application of CBR. The functions of the components of the system are illustrated by examples. Finally, some existing case-based schedulers are compared with CBS-1.

INTRODUCTION

Scheduling is the allocation of resources, like machines or human power, to operations over time to achieve certain goals. In job-shop scheduling the goals to be achieved are the processing or production of discrete parts in several steps each requiring several different resources. Dynamic scheduling is scheduling simultaneously with the execution of the processes that are affected by the created schedules.

In the Interuniversitary Centre for CIM (IUCCIM) in Vienna the production process for remote controlled toy cars is used to demonstrate the main ideas in CIM. In this context the problem of scheduling incoming orders for toy cars into the ongoing production process arises. There are several reasons for the complexity of such a scheduling task.

• There is a combinatorial explosion of the number of possible schedules (which must be checked for feasibility) in each problem dimension such as the number of machines and operations.

• […]

A number of practical knowledge acquisition methodologies and tools have been based on the elicitation and analysis of repertory grids. These result in frames and rules that are exported to knowledge-based system shells. In the development of repertory grid tools, the original methodology has been greatly extended to encompass the data types required in knowledge-based systems. However, this has been done on a fairly pragmatic basis, and it has not been clear how the resultant knowledge acquisition systems relate to psychological, or computational, theories of knowledge representation. This paper shows that there is a close correspondence between the intensional logics of knowledge, belief and action developed in the personal construct psychology underlying repertory grids, and the intensional logics for term subsumption knowledge representation underlying KL-ONE-like systems. The paper gives an overview of personal construct psychology and its expression as an intensional logic describing the cognitive processes of anticipatory agents, and uses this to survey knowledge acquisition tools deriving from personal construct psychology.

PERSONAL CONSTRUCT PSYCHOLOGY

George Kelly was a clinical psychologist who lived between 1905 and 1967, published a two volume work (Kelly, 1955) defining personal construct psychology in 1955, and went on to publish a large number of papers further developing the theory many of which have been issued in collected form (Maher, 1969). Kelly was a keen geometer with experience in navigation and an interest in multi-dimensional geometry.

Abstract

We describe a method for Multi-Agent System design which is assisted by two original typologies, resulting from the deeper study of knowledge and reasoning. The first typology reflects a formal character while the second reflects a technological character. The purpose of the Formal Typology is the classification and structuring of knowledge and reasoning. The Technological Typology handles the parameters governing the reasoning intrinsic to Multi-Agent technology, not only at the individual level of the agent but also within a group of agents. Possible correspondence between both of these typologies will become concrete by the presentation of the Multi-Agent generator MAPS (Multi-Agent Problem Solver), and the Multi-Agent system KIDS (Knowledge based Image Diagnosis System) devoted to Biomedical Image Interpretation.

Keywords

Second Generation Expert Systems, Multi-Agent System Design, Distributed Artificial Intelligence, Knowledge and Reasoning Modeling, Control, Biomedical Image Interpretation.

INTRODUCTION

Among knowledge based systems using artificial intelligence techniques we are particularly interested in the Multi-Agent systems which arise from their second generation (systems using multiple reasoning schemes). The Multi-Agent paradigm results from distributed artificial intelligence approaches and makes it possible to overcome the drawbacks encountered during the resolution of complex problems. The main issue of the Multi-Agent approach involves the distribution of tasks and skills among intelligent entities that co-operate, pooling their knowledge and their expertise to attain an aim (Ferber 88). In this way, not only a multi-modal knowledge representation, and reasoning schemes handling are permitted but also, co-operative problem solving.

Introduction

Time is an important aspect of any intelligent knowledge representation. This has led to a rising need for reasoning about time in various applications of artificial intelligence such as process control or decision making. Different schemes for temporal information representation have been proposed so far. A natural way to refer to a temporal event consists in making references to a clock providing a quantitative or numerical representation of time, as well as several concepts such as duration and calendar. However, a clock reference is not always available or relevant. In such cases, a qualitative (symbolic) representation of time can be used to describe the situations in question.

In spite of the different representations of temporal information proposed, most are not completely satisfactory. Looking at existing work, Allen's representation [Allen 83] is a very powerful representation in describing the relativity between intervals. Vilain and kautz proposed a subinterval algebra [Vilain & Kautz 86], Ghallab and Mounir [Ghallab & Mounir 89] based on the notion of subinterval algebara, have also proposed a model with symbolic relations but within the framework of subinterval algebra. However, these models don't address numerical aspect of time. On the other hand, the time map of Dean and McDermott [Dean & McDermott 87], Rit geometrical model [Rit 86] and the temporal constraint networks [Dechter, Meiri & Pearl 91] are designed for handling metric information and can't handle in a good way symbolic ones.

INTRODUCTION

Examples of the pattern classification problem (known variously as: pattern recognition, discriminant analysis, and pattern grouping) are widespread. In general such problems involve the need to assign objects to various groups, or classes, and include such applications as: (i) the assignment of production items to either defective or non-defective classes as based upon the results of tests performed on each part, (ii) the assignment of personnel to jobs as based upon their test scores and/or physical attributes, (iii) the assignment of an object detected by radar to either a friendly or unfriendly category, (iv) the categorization of investment opportunities into those that are attractive and those that are not, and so on. Early (scientific) efforts to model and solve the pattern classification problem utilized, for the most part, statistical approaches. In turn, these approaches usually rely upon the somewhat restrictive assumptions of multivariate normal distributions and certain types of (and conditions on) covariance matrices. More recent attempts have employed expert systems, linear programming (LP) and, in particular, neural networks. In this paper, we describe the development of an approach that combines linear programming (specifically, traditional linear programming and/or linear goal programming [Ignizio, 1982]) with neural networks, wherein the combined technique is itself monitored and controlled by an expert systems interface.

More specifically, we describe the use of expert systems and linear programming in the simultaneous design and training of neural networks for the pattern classification problem.

Abstract

The development of expert systems is inherently uncertain and so involves a high degree of risk. This paper describes a project management method that helps manage this uncertainty. It has been tailored to the Client Centred Approach — an expert system development method that is being designed for use by small and medium sized enterprises. This context implies that the management technique and its accompanying documentation must not over burden the resources of a smaller developer. The helix method of project management introduced in this paper represents a different view of Boehm's Spiral Model. It accepts that conventional linear project planning methods are not always suitable for developers of expert systems. Having accepted this, the helix method allows plans to be made for each development stage within the Client Centred Approach. We believe the Client Centred Approach is applicable wherever prototyping is used, and we contrast it with the methods being developed by KADS-II.

INTRODUCTION

This paper describes proposals for handling project management within the Client Centred Approach (CCA). The principles of the CCA are described in Basden [1989]. The thinking behind the approach, and its current state of development, are described in greater detail in Basden et al. [1991] and Watson et al. [1992]. Although the technique described here is applicable for any project that uses prototyping to develop a system (around forty five per cent of all commercial expert system projects according to a recent survey [DTI, 1992]), it has been developed specifically for small and medium sized enterprises (SMEs), rather than larger organisations.

INTRODUCTION

The last few years has seen a significant change in commercial KBS development. Organisations are now building KBS to solve specific business problems rather than simply to see what the technology can do. There has also been a move away from building KBS on stand alone PCs to using the corporate resources of networks, mini and Mainframe computers, and existing databases. As a result of these changes, two significant questions are now being regularly asked by organisations developing or interested in developing KBS:

• How can KBS be linked into existing systems to enhance their processing functions and make better use of data already held?

• What methods can be used to help build commercial applications using KBS techniques?

The key to these questions is the use of an integrated approach to the development of all IT systems. There are many methods available for conventional systems development, such as Information Engineering, SSADM, Jackson and Yourdon. There are also a number of KBS methods available or under development such as KADS, KEATS, and GEMINI. However, commercial organisations with well established procedures for conventional development do not want to use two different methods side-by-side, nor do they wish to discard their current conventional development method and replace it with a method claiming to cover all aspects of conventional and KBS development. Organisations therefore require some way of integrating KBS methods into their existing methods.

Abstract: Current intelligent user interfaces have two limitations: (i) They are domain specific and mainly built for existing database management systems, (ii) They are specific to the target systems for which they are constructed. However, user goals, which motivate interactions with a computer, are likely to be complicated and to require the use of multiple target systems in various domains. In this paper, we discuss the development of intelligent user interfaces which are not subject to the limitations identified. An architecture is proposed, the major function of which is the dynamic integration and intelligent use of multiple target systems relevant to a user's goals. Other important features of the proposed system include its theoretical orientation around relevance relationships, mental models and speech acts, and the introduction of “system experts” and “goal manager”. A prototype Intelligent Multifunctional User Interface, (IMUI), is briefly described which indicates that the proposed architecture is viable, the methodology is promising, and the theoretical ideas introduced are worthy of further investigation.

INTRODUCTION

Computer-based systems are coming to play an ever more important part in our society, and as they do so, they become increasingly complicated and difficult to use effectively. As a consequence, the need to develop flexible and versatile intelligent interfaces has become more crucial than ever.

What would an ideal interface look like, and how can such a system be designed and implemented? Most investigators would agree that it should behave like an intelligent human assistant who has expert knowledge both of user characteristics and requirements, and of target system(s).

Abstract

This paper describes and illustrates the use of a methodology suitable for the formal development of expert systems. It addresses the problems of verification and validation of expert systems in a realistic way, though the methods are not advocated as a general tool for expert system development. The framework described allows for both the specification of Knowledge and the specification of the Inference methods which provide the basis for deduction. A flexible and extensible environment for the development and testing of specific types of expert system is presented. Various tools and results are shown to be useful in determining properties of both the knowledge base and the inference system when these are developed within the proposed framework.

The framework is based on exploitation of the transformational model of software development in combination with techniques from algebraic specification.

INTRODUCTION

The development of expert systems, within a formal development framework (see [Krieg-Brückner and Hoffmann 91]), can be seen as a significant advance in expert system technology. The benefits accrued from such an approach are substantial. In particular the following are notable: a formal foundation for reasoning about properties of the knowledge base and inference system is provided. Inductive and deductive methods are available to help in both the construction of the expert system and as a tool for analysis of the knowledge bases. A well defined language, with well defined semantics, provides the basis for specifying both the expert system and the associated knowledge bases.

Abstract

One of the principal difficulties in developing a distributed problem solver is how to distribute the reasoning task between the agents cooperating to find a solution.

We will propose the distributed logic programming language DLP as a vehicle for the design and implementation of distributed knowledge based systems. The language DLP combines logic programming with active objects.

We will show how object oriented modeling may be applied for the specification and implementation of a distributed diagnostic (medical) expert system. The example illustrates how the diagnostic process is distributed over the agents participating in the diagnosis according to the structure of the knowledge of that particular domain.

Keywords:object oriented modeling, distributed problem-solving, knowledge-based systems, expert systems, distributed logic programming

Introduction

Logic programming offers a declarative way to solve problems in Artificial Intelligence. However, when implementing large (possibly distributed) systems, traditional software engineering problems such as modularization and the distribution of data and control reoccur. Cf. [Subrahmanyam, 1985].

One of the principal difficulties in developing a distributed problem solver is how to distribute the reasoning task between the agents cooperating to find a solution.

Due to its declarative nature, logic programming has become popular for implementing knowledge-based systems. However, lacking adequate modularization facilities, logic programming languages such as Prolog fall short in providing the mechanisms necessary to specify the distribution of data and control.

Object oriented modeling To tackle these problems, we suggest in this paper to embed the logic programming paradigm into an object oriented approach.

Abstract

Most expert systems perform a task on behalf of the user. The task usually involves gathering and analyzing data, and recommending or initiating the appropriate action. However, expert systems can also play an important role in showing the user how to perform a task. In this role, the expert system provides support until it eventually becomes of decreasing importance as its knowledge base is transferred to the user. This category includes Help Systems, Coaching Systems, and Tutorial Systems. In this paper, we discuss the development of an Intelligent Advisor combining the three functions in a system to assist the user in acquiring and refining the knowledge required to carry out a design task. The combined system provides a means of introducing a training facility as an integral part of the work environment. The primary goal of our project is the creation of a system in which the generic advisor components are identified along with the methodology required to adapt them to specific applications. The conceptual modelling phase of database design was chosen as the application domain to develop the system and to demonstrate feasibility. An initial prototype has been implemented, which illustrates the operation of the system in each of the three modes as applied to database modelling. The technology is currently being extended to a second application domain.

Introduction

ERMA (Entity-Relationship Modelling Advisor) is a knowledge-based system that serves as a consultant to the user of a computer-based design tool, providing advice as required.

Abstract

The Machine Learning Toolbox (MLT), an Esprit project (P2154), provides an integrated toolbox of ten Machine Learning (ML) algorithms. One distinct component of the toolbox is Consultant, an advice-giving expert system, which assists a domain expert to choose and use a suitable algorithm for his learning problem. The University of Aberdeen has been responsible for the design and implementation of Consultant.

Consultant's knowledge and domain is unusual in several respects. Its knowledge represents the integrated expertise of ten algorithm developers, whose algorithms offer a range of ML techniques; but also some algorithms use fairly similar approaches. The lack of an agreed ML terminology was the initial impetus for an extensive, associated help system. From an MLT user's point of view, an ML beginner requires significant assistance with terminology and techniques, and can benefit from having access to previous, successful applications of ML to similar problems; but in contrast a more experienced user of ML does not wish constant supervision. This paper describes Consultant, discusses the methods used to achieve the required flexibility of use, and compares Consultant's similarities and distinguishing features with more standard expert system applications.

INTRODUCTION

The Machine Learning Toolbox (MLT), an Esprit project (P2154), provides an integrated toolbox of ten Machine Learning (ML) algorithms. One distinct component of the toolbox is Consultant, an advice-giving expert system. It provides domain experts with assistance and guidance on the selection and use of tools from the toolbox, but it is specifically aimed at experts who are not familiar with ML and its design has focused on their needs.

Abstract

An intelligent learning data base (ILDB) system is an integrated learning system which implements automatic knowledge acquisition from data bases by providing formalisms for 1) translation of standard data base information into a form suitable for use by its induction engines. 2) using induction techniques to produce knowledge from data bases, and 3) interpreting the knowledge produced efficiently to solve users' problems. Although a lot of work on knowledge acquisition from data bases has been done, the requirements for building practical learning systems to learn from conventional data bases are still far away for existing systems to reach. A crucial requirement is more efficient learning algorithms as realistic data bases are usually fairly large. Based on KEshell. dBASE3 and the low-order polynomial induction algorithm HCV. this paper presents a knowledge engineering shell. KEsheH2. which implements the 3 phases of automatic knowledge acquisition from data bases in an integral way.

INTRODUCTION

Over the past twenty years data base research has evolved technologies that are now widely used in almost every computing and scientific field. However, many new advanced applications including computer-aided design (CAD), computer-aided manufacturing (CAM), computer-aided software engineering (CASE), image processing, and office automation (OA) have revealed that traditional data base management systems (DBMSs) are inadequate, especially on the following cases [Wu 90b]:

• Conventional data base technology has laid particular stress on dealing with large amounts of persistent and highly structured data efficiently and using transactions for concurrency control and recovery.

• […]

Abstract

This paper describes work on the construction of a configurable knowledge acquisition tool, Jigsaw. Unlike automated knowledge acquisition programs such as MORE [Kahn, 1988], MOLE [Eshelman, 1988], and OPAL [Musen, 1989], each of which automates elicitation for just one problem solving method, it is possible to alter Jigsaw's knowledge acquisition strategy to match different problem solving methods.

The work is based upon eliciting knowledge for problem solvers made up from different combinations of generic task (as denned in [Chandrasekaran, 1986] and [Chandrasekaran, 1988]). Each combination of generic tasks defines the functionality of a different problem solving method. However, the eventual aim of this work is that it will be possible to adapt it to a range of different KADS [Schreiber et. al., 1987] interpretation models and thus it will be part of a complete knowledge acquisition methodology.

The paper outlines the requirements for such a knowledge acquisition tool and details the distributed architecture which allows the tool, Jigsaw, to achieve the required flexibility to elicit knowledge for such problem solvers. An important part of this flexibility is the way in which Jigsaw can be configured to match different types of problem solver. This is described in some detail.

Jigsaw has been used to reproduce the MDX2 [Sticklen, 1987] knowledge base, which was initially constructed by using manual knowledge acquisition techniques. The paper gives a description of how Jigsaw elicited this knowledge.

The routing algorithms in Chapter 5 can be converted into efficient asynchronous algorithms by replacing the global clock with a synchronization scheme based on message passing. We also demonstrate that asynchronous fsra has low sensitivity to variations in link and processor speeds.

Introduction

The assumption of synchronism often greatly simplifies the design of algorithms, be they sequential or parallel. Many computation models — for example, RAM (“Random Access Machine”) [5] in the sequential setting and PRAM in the parallel setting — assume the existence of a global clock. But, this assumption will become less desirable as the number of processors increases. For one thing, a global clock introduces a single point of failure. A global clock also restrains each processor's degree of autonomy and renders the machine unable to exploit differences in running speed [42, 192], limiting the overall speed to, so to speak, that of the (“slowest” component instead of the “average” one, thus wasting cycles. Tight synchronization also limits the size of the parallel computer, since it takes time to distribute the clock-signal to the whole system [316].

Proceeding in epochs, our routing schemes in Chapter 5 assume synchronism. In fact, the very definition of ECS assumes a global clock to synchronize epochs. We show in this chapter that with synchronization done via message passing, ECSs can be made asynchronous without loss of efficiency and without global control. Much work has been done in this area; see, for example, [31, 32, 33].