XPLAIN: A System for Creating and Explaining Expert Consulting Programs

By William R. Swartout. 1983

Traditional methods for explaining programs provide explanations by converting the code of the program or traces of its execution to English. While such methods can sometimes adequately explain program behavior, they typically cannot provide justifications for that behavior. That is, such systems cannot tell why what the system is doing is a reasonable thing to be doing. The problem is that the knowledge required to provide these justifications was used to produce the program but is itself not recorded as part of the code, and hence is unavailable. The XPLAIN system uses an automatic programmer to generate a consulting program by refinement from abstract goals. The automatic programmer uses a domain model, consisting of descriptive facts about the application domain, and a set of domain principles which prescribe behavior and drive the refinement process forward. By examining the refinement structure created by the automatic programmer, XPLAIN provides justifications of the code. XPLAIN has been used to re-implement major portions of a Digitalis Therapy Advisor and provides superior explanations of its behavior.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Explanation in Human-AI Systems: A Literature Meta-Review, Synopsis of Key Ideas and Publications, and Bibliography for Explainable AI (2019, methodology): “…Scott, Clancey, Davis, and Shortliffe, 1977; Scott, Clancey, Davis, and Shortliffe, 1984; Shortliffe, 1976; Silverman, 1992; Southwick, 1988, 1988, Swartout, 1977, 1981, 1983, 1985; Swartout and Moore, 1993; Swartout, Paris, and Moore, 1991; Wallis and Shortliffe, 1981, 1984; Weiner, 1980; Wick…”
The challenge of crafting intelligible intelligence (2018, background): “n practice in AI systems. For example, early rule-based expert systems included explanation systems that augmented a trace of the system’s reasoning — on a particular case — with background knowledge [38]. Recommender systems, which were one of the first deployed uses of machine learning, also induc”
Explanation and Argumentation Capabilities:Towards the Creation of More Persuasive Agents (2002, background): “The xplain system (Swartout 1983; Neches et al. 1985; Swartout et al. 1991) provides deep explanations by using the trace that an automatic programmer leaves behind and a high-level knowledge base representing domain facts and problem-solving plans.”
Explaining reasoning: an overview of explanation in knowledge-based systems (1991, methodology): “In XPLAIN (Swartout, 1983) the computer artifact problem is handled by attaching a viewpoint to steps in a domain principle.”
Goal-Based Explanation Evaluation (1991, background): “their decision paths, but the reasoning underlying those paths ( Swartout, 1983 ); systems to devise”
Interacting meaningfully with machine learning systems: Three experiments (2009, background): “The bulk of the literature on explaining intelligent systems has focused on explaining expert systems based on deductive logic or rules (Swartout 1983; Clancey 1983; Wick and Thompson 1992; Johnson 1994).”
Explanatory Power of Intelligent Systems : A Research Framework (2004, unknown): “As Swartout (1983) astutely observes in discussing the MYCIN rule traces: " Parts of the program [i.e., rule traces] appear mainly because we are implementing an algorithm on a computer.”
The Use and Effects of Knowledge-Based System Explanations: Theoretical Foundations and a Framework for Empirical Evaluation (1996, background): “, explanation-based learning), and 2) the idea that knowledge can be optimally structured or rep resented in a knowledge-base to facilitate the provision of explanations to users (Swartout 1983).”
Graphical Explanation in Belief Networks (1997, background): “Swartout (1983) comments that Otrust in a system is developed not only by the quality of the results but also by clear description of how they were derived...”
Trends and Trajectories for Explainable, Accountable and Intelligible Systems: An HCI Research Agenda (2018, background): “Historically, there has been occasional interest in explanations of intelligent systems over the past decades with expert systems in the 1970s [165, 172], Bayesian networks (for a review, refer to [102]) and artificial neural networks [6] in the 1980s, and recommender systems in the 2000s [33, 81].”