Explaining quantitative systems to uninitiated users

The importance of explanation in expert systems has been documented from the early days of their development; there is an equally pressing need for explanation in systems that employ a decision-making process based on quantitative reasoning. This is particularly necessary for users who do not have a sophisticated understanding of the formal apparatus that the system employs to reach its decisions. In order to generate meaningful answers to questions asked by such unsophisticated users, an explanation facility must translate the formal structures of the problem solving system into the concepts with which the user understands the problem domain. Previous work on the explanation of quantitative systems is based on the assumption that the user has at least a basic grasp of the formal approach of the problem solving system. However, in realistic application situations, it is more likely the case that in order for the human user to understand why a mathematically-based advice-giving system makes the suggestions that it does, the problem solving rationale of the system must be explained in the user's own terms, which are typically different from those of the mathematical system. To develop an explanation methodology that is capable of justifying the results of a system based on quantitative reasoning to an uninitiated user, we employ a representation that enables our explanation facility to translate the abstract mathematical relationships that make up a quantitative system into the domain-specific concepts with which a typical user approaches the problem solving task. In our system, the process of generating explanations, therefore, involves translating one set of concepts into another. An added feature of this system is that it is capable of providing explanations from two perspectives: that of the quantitative problem solving system, and that of the human user who is familiar with the domain problem but not with the mathematical approach. We have implemented this approach to explaining quantitative systems by creating an explanation facility for a problem in the manufacturing domain. This facility responds to user queries about a scheduling system that uses a mathematically-based heuristic to choose jobs for an annealing furnace.     

On making expert systems more like experts

Abstract: Expert systems still lack the skill of an expert when it comes to providing explanations of the results of expert reasoning. This is because while such systems may implement knowledge which is sufficient to mimic the performance of an expert, they do not necessarily model the expertise upon which that performance is based. Such a model must include knowledge of that domain's terminology, knowledge of domain facts, and knowledge of problem-solving methods. The Explainable Expert Systems project has been exploring a new paradigm for expert system development that is intended to capture such missing knowledge and make it available for explanation. This paper will discuss the principles behind this paradigm and consider two systems that employ it.     

A systematic review and taxonomy of explanations in decision support and recommender systems

With the recent advances in the field of artificial intelligence, an increasing number of decision-making tasks are delegated to software systems. A key requirement for the success and adoption of such systems is that users must trust system choices or even fully automated decisions. To achieve this, explanation facilities have been widely investigated as a means of establishing trust in these systems since the early years of expert systems. With today’s increasingly sophisticated machine learning algorithms, new challenges in the context of explanations, accountability, and trust towards such systems constantly arise. In this work, we systematically review the literature on explanations in advice-giving systems. This is a family of systems that includes recommender systems, which is one of the most successful classes of advice-giving software in practice. We investigate the purposes of explanations as well as how they are generated, presented to users, and evaluated. As a result, we derive a novel comprehensive taxonomy of aspects to be considered when designing explanation facilities for current and future decision support systems. The taxonomy includes a variety of different facets, such as explanation objective, responsiveness, content and presentation. Moreover, we identified several challenges that remain unaddressed so far, for example related to fine-grained issues associated with the presentation of explanations and how explanation facilities are evaluated.     

A structural equation modeling approach to generate explanations for induced rules

The performance of an expert system depends on the quality and validity of the domain-specific knowledge built into the system. In most cases, however, domain knowledge (e.g. stock market behavior knowledge) is unstructured and differs from one domain expert to another. So, in order to acquire domain knowledge, expert system developers often take an induction approach in which a set of general rules is constructed from past examples. Expert systems based upon the induced rules were reported to perform quite well in the hold-out sample test.

However, these systems hardly provide users with an explanation which would clarify the results of a reasoning process. For this reason, users would remain unsure about whether to accept the system conclusion or not. This paper presents an approach in which explanations about the induced rules are constructed. Our approach applies the structural equation model to the quantitative data, the qualitative format of which was originally used in rule induction. This approach was implemented with Korean stock market data to show that a plausible explanation about the induced rule can be constructed.     

On two AI traditions

To understand the role of expert systems as a medium for transferring knowledge and skills within organisations requires an understanding of the nature of expertise within working life contexts. Central to this issue of transfer is the debate on the nature of tacit/implicit knowledge and the problem of formalising it in explicit form. This paper considers the British approach to the development of knowledge-based systems, which is regarded as being predominantly rationalistic, and compares it with the Scandinavian approach, which is regarded as being predominantly humanistic. The paper proposes that crucial to the debate on the transfer of knowledge and skills is the development of expert systems as communications media which aims at enhancing and sharing the knowledge and skills of users. A human-centred approach for the future is proposed which brings the above two traditions together thereby providing a developmental framework for knowledge based systems.     

Formalizing dialectical explanation support for argument-based reasoning in knowledge-based systems

The concept of explanation has received attention from different areas in Computer Science, particularly in the knowledge-based systems and expert systems communities. At the same time, argumentation has evolved as a new paradigm for conceptualizing commonsense reasoning, resulting in the formalization of different argumentation frameworks and the development of several real-world argument-based applications. Although the notions of explanation and argument for a claim share many common elements in knowledge-based systems their interrelationships have not yet been formally studied in the context of the current argumentation research in Artificial Intelligence. This article explores these ideas by providing a new perspective on how to formalize dialectical explanation support for argument-based reasoning. To do this, we propose a formalization of explanations for abstract argumentation frameworks with dialectical constraints where different emerging properties are studied and analyzed. As a concrete example of the formalism introduced we show how it can be fleshed out in an implemented rule-based argumentation system.     

Folksonomy‐Based Recommender Systems: A State‐of‐the‐Art Review

Collaborative tagging systems, also known as folksonomies, have grown in popularity over the Web on account of their simplicity to organize several types of content (e.g., Web pages, pictures, and video) using open‐ended tags. The rapid adoption of these systems has led to an increasing amount of users providing information about themselves and, at the same time, a growing and rich corpus of social knowledge that can be exploited by recommendation technologies. In this context, tripartite relationships between users, resources, and tags contained in folksonomies set new challenges for knowledge discovery approaches to be applied for the purposes of assisting users through recommendation systems. This review aims at providing a comprehensive overview of the literature in the field of folksonomy‐based recommender systems. Current recommendation approaches stemming from fields such as user modeling, collaborative filtering, content, and link‐analysis are reviewed and discussed to provide a starting point for researchers in the field as well as explore future research lines.     

Interactive Case-Based Reasoning in Sequential Diagnosis

Interactive trouble-shooting and customer help-desk support, both activities that involve sequential diagnosis, represent the majority of applications of case-based reasoning (CBR). An analysis is presented of the user-interface requirements of intelligent systems for sequential diagnosis. We argue that mixed-initiative dialogue, explanation of reasoning, and sensitivity analysis are essential to meet the needs of experienced as well as novice users. Other issues to be addressed by system designers include relevance and consistency in dialogue, tolerance of missing data, and timely provision of feedback to users. Many of these issues have previously been addressed by the developers of expert systems and the lessons learned may have important implications for CBR. We present a prototype environment for interactive CBR in sequential diagnosis, called CBR Strategist, which is designed to meet the identified requirements.     

Human operators and real-time expert systems

Abstract: This paper describes and examines real-time expert systems from the perspective of their users. It categorizes real-time expert systems according to a three-component system consisting of an expert system, a user and a process being controlled, by considering all possible arrangements of information flow between the components, and uses this classification as a basis for reviewing real-time expert systems. The focus of interest lies in those characteristics of real-time expert systems which impinge on users, especially as they have been discussed by computer scientists. After critically examining these claims, the paper discusses the interaction between systems and users at the perceptual/motor, cognitive and supervisory levels. It concludes by arguing that the successful design and installation of real-time expert systems require the application of ergonomics techniques to provide for efficient and accurate user-system interaction.     

The quality of expert probability judgement: issues and analysis

Abstract: Many decision-aiding technologies require valid probability judgements to be elicited from domain experts. But how valid are experts' probability judgements? We describe two approaches to the assessment of quality of probability judgement—calibration and coherence—and review the research findings following from these two approaches. In many cases, expert probability judgement has been found to lack validity and this sub-optimality has largely been attributed to computational errors on the part of the expert. The preferred solution to poor validity in probability judgement has therefore been to reduce the amount of computation performed by the expert. Complex probabilities can be calculated mechanically from simple probability judgements elicited from the expert. We present evidence which suggests that this recomposition technique doesn't guarantee valid probabilities. Our explanation for this finding is that there are various problems concerned with eliciting even the simple probabilities which are necessary for subsequent recomposition. We conclude by proposing some solutions to these elicitation problems which should help ensure that probability judgements of increased validity are available to those attempting to capture subjective assessments for input into decision support systems.     

Integrating artificial neural networks with rule-based expert systems

The Rule-Based (RB) and the Artificial Neural Network (ANN) approaches to expert systems development have each demonstrated some specific advantages and disadvantages. These two approaches can be integrated to exploit the advantages and minimize the disadvantages of each method used alone. An RB/ANN integrated approach is proposed to facilitate the development of an expert system which provides a “high-performance” knowledge-based network, an explanation facility, and an input/output facility. In this case study an expert system designed to assist managers in forecasting the performance of stock prices is developed to demonstrate the advantages of this integrated approach and how it can enhance support for managerial decision making.     

Cooperative problem solving and explanation

Recent studies have pointed out several limitations of expert systems regarding user needs, and have introduced the concepts of cooperation and joint cognitive systems. While research on explanation generation by expert systems has been widely developed, there has been little consideration of explanation in relation to cooperative systems. Our aim is to elaborate a conceptual framework for studying explanation in cooperation. This work relies heavily on the study of human-human cooperative dialogues. We present our results according to two dimensions; namely, the relation between explanation and problem solving, and the explanation process itself. Finally, we discuss the implications of these results for the design of cooperative systems.     

Knowledge-Based System Explanation: The Ripple-Down Rules Alternative

The ability to provide explanations has been seen as a key feature of expert systems (ES) typically not offered by other types ofcomputer systems. ES need to offer explanations because ofimprecise domains and the use ofheuristics. Verification is not enough. ES need to justify and be accountable. Explanation is seen as an important activity for knowledge-based systems as it satisfies the user's need to decide whether to accept or reject a recommendation. In this paper we review explanation in first-generation and second-generation ES. An alternative is offered to the main approaches which uses multiple classification ripple-down rules and challenges even the goals of explanation. Instead of trying to give explanations which provide a meaningful line of reasoning and which are tailored to suit the individual it may be just as useful to provide the user with sufficient information and browsing tools to develop their own line of reasoning. The type of information that can assist understanding is the context in which the recommendation applies (which is provided through the display ofrelevant cases and exception rule history) and the ability to explore an abstraction hierarchy of the rules using formal concept analysis. An explanation tool kit aimed at putting the user in control is described and evaluated in this paper. Received 15 January 2001 / Revised 21 June 2001 / Accepted in revised form 1 October 2001     

A knowledge-based system for postgraduate engineering courses

This paper describes a novel approach to the development and implementation of an educational tool based on knowledge-based technology employing an expert system shell. Software has been developed which, after providing basic background information, proceeds to analyse the student's learning pattern to establish which next question, explanation, or topic to propound in order to aid the learning process and ensure that fundamental knowledge is gained by the student at his/her own pace. While the system described has been designed and implemented specifically to supplement, rather than supplant, the taught part of an MSc. course in Manufacturing Systems Engineering and Management, the methodology formulated can be used to develop similar knowledge-based systems for other technical as well as non-technical subjects at both undergraduate and postgraduate levels in Higher Education.     

TOWARD A NATURAL LANGUAGE-BASED CAUSAL MODEL ACQUISITION SYSTEM

Future expert systems for understanding physical mechanisms will probably employ causal models as the foundation of their expertise, and the problem of acquiring these causal models is important. This paper explores one possibility, that of acquiring causal models by understanding natural language explanations of these mechanisms. It identifies six different research issues in which understanding an explanation requires knowledge-based reasoning, and proposes approaches to these problems within an integrated natural language-based causal model acquisition system.     

Expert systems in manufacturing. Part 2: A management framework for expert-systems innovation and technology transfer

Current approaches to expert systems technology transfer have tended to focus upon the marketing and servicing of technology capabilities and potential whilst remaining uncertain about the process factors which determine how this technology may be applied and adopted effectively. Furthermore, much of current expert systems research work and literature addresses these issues from the viewpoint of the supplier or donor whilst overlooking the importance of human and organisational perspectives which shed light on the means of delivery and take-up within the recipient organisation. The paper, the second of two that look at expert systems innovation in manufacturing, argues for greater consideration of the characteristics, processes and mechanisms of technology transfer. It defines a new conceptual basis for technology transfer which stresses a ‘needs-driven’ process of change; this highlights the importance of context as well as content in expert systems transfer and implementation. From this, a management framework is outlined and is used to rationalise the transfer problems and needs described in the first paper following a survey of 145 manufacturing users. It is also shown how this framework may be used to understand more about the multi-level and multi-dimensional needs and effects of technology induced change and therefore how it may be used to help senior management strategically plan and co-ordinate expert systems programmes in their organisations.     

Applying expert principles to computer systems development: Using compilers and generators

Expert systems can be used for developing computer systems. The expert system can help in expressing the definition of the problem, as well as answering the problem. So it deals with ‘how’ and ‘what’ in the development of a computer system. The two main approaches to providing solution expertise are to use a compiler or a generator. The generator approach is more limited than the compiler approach. The Systematics generator includes precoded design templates in its knowledge base. The two main approaches to problem statement are data-driven or goal-driven.     

Article Expertext: Intelligent hypertext

Abstract

Intelligent tutoring media can take advantage of hypertext systems and expert systems. Hypertext is text plus an abstraction of the text. In reasoning with this abstraction, a hypertext system manifests expertise. This intelligent hypertext is called expertext. When the abstraction of the text is a semantic network, spreading activation and inheritance methods may help people appreciate the structure of the hypertext. When predicates and implications are included in the abstraction, logical deductions can be performed. Queries can be logically answered and then a sequence of relevant text blocks can be provided as an explanation of the answer. Distributed expertext systems use knowledge bases about the domain of the text and about the users. Distributed expertext systems have been developed which guide people in searching for information and in creating information.     

ACTIVE vs. PASSIVE APPROACHES TO INTELLIGENT PROGRAM DIAGNOSIS

Much of research in intelligent programming systems for users has been polarized towards two opposite domains: active and passive approaches to diagnosis. The advocates of the active approach claim that much of the effectiveness of intelligent program systems is contributed to having strong control over the behavior of the users and providing immediate feedback on errors and misconceptions. Opponents of this approach, on the other hand, have argued that active approach through its interventionist style does not provide users the flexibility needed to observe their own behavior and discover their own errors, hence the users are not given an opportunity to selfdebug their solutions. This paper covers the engineering of intelligent program diagnosis systems and reports an empirical evaluation which attempts to get some insights into the superiority of active approach over passive approach or vice versa. The evaluation is conducted using our prototype system DISCOVER. The system provides a visualization-based environment which supports both active and passive modes of intelligent program diagnosis.