This paper attempts to assess the practical utility of nonmonotonic logic in diagnostic problem solving. We begin with a brief review of the main assumptions which motivate work in this area, and discuss two logic-based approaches which involve nonmonotonic arguments. Then we consider two recent proposals for the application of default logic to diagnosis, as well as a proposal based on counterfactual logic. In conclusion, we briefly compare these methods with other diagnostic reasoning paradigms found in the Artificial Intelligence literature.

Computational modelling of scientific discovery is emerging as an important research field in artificial intelligence. Various computational systems modelling different aspects of scientific research and discovery have been developed. This paper looks at some of these models in order to examine how knowledge is organized in such systems, what forms of representation they have, how their methods of learning and representation are integrated, and the effects of representation on learning. The paper also describes the achievements and shortcomings of these systems, and discusses the obstacles in developing more comprehensive models.

This work studies quantitative measures for ranking judicial decisions by the Brazilian Supreme Court [Supremo Tribunal Federal (STF)] and selecting leading cases, which are understood as those with broadness of influence on different legal fields. The measures are based on a network built over decisions whose cases were finalized in the Brazilian Supreme Court between 01/2001 and 12/2019, which were obtained by crawling publicly available STF records. Three ranking measures are proposed; two are adaptations of the PageRank algorithm, and one adapts Kleinberg’s algorithm. Such measures are compared with respect to agreement on top 100 rankings; we also analyze each robustness measure based on self-agreement under perturbation.

We examine whether the resulting quantitative ranking is congenial to a qualitative intuition of what the legal community usually considers as relevant precedents. We also discuss some possible criteria of relevance in the seek for patterns that suggest how quantitative and qualitative measures would better align. The ranking of leading cases and relevant decisions improved after building decision networks without irrelevant appeals and decisions that overflow the court offers a starting point to discuss the role of STF in the Brazilian judicial system.

In our last work, both versions of PageRank and Kleinberg algorithms produced different rankings and all of them were robust with respect to 10% and 20%-perturbation levels, but none of them retrieved leading cases at the top of these rankings. Then, we took a further step in the studies of the STF decision network and we introduced better filtering of network nodes guided by legal expertise on the works of the Supreme Court. We also introduced more fine-grained perturbance levels to understand the impact of such filters in the STF decision network. We concluded that after filtering low-relevance decision types, the STF decision network is still robust under 10%-perturbation, but it presents higher degradation by increasing perturbation levels. The two versions of PageRank still produce different rankings. Kleinberg’s algorithm provides a different ranking, with many relevant criminal cases. Although we improved algorithms rankings filtering decisions from the network, which represents an important methodological step, there is still room for improvement. Given that relevant decisions are well ranked after filtering out a large amount of irrelevant decisions, the results set a starting point to discuss the role of STF in the Brazilian judicial system.

One challenge faced by reinforcement learning (RL) agents is that in many environments the reward signal is sparse, leading to slow improvement of the agent’s performance in early learning episodes. Potential-based reward shaping can help to resolve the aforementioned issue of sparse reward by incorporating an expert’s domain knowledge into the learning through a potential function. Past work on reinforcement learning from demonstration (RLfD) directly mapped (sub-optimal) human expert demonstration to a potential function, which can speed up RL. In this paper we propose an introspective RL agent that significantly further speeds up the learning. An introspective RL agent records its state–action decisions and experience during learning in a priority queue. Good quality decisions, according to a Monte Carlo estimation, will be kept in the queue, while poorer decisions will be rejected. The queue is then used as demonstration to speed up RL via reward shaping. A human expert’s demonstration can be used to initialize the priority queue before the learning process starts. Experimental validation in the 4-dimensional CartPole domain and the 27-dimensional Super Mario AI domain shows that our approach significantly outperforms non-introspective RL and state-of-the-art approaches in RLfD in both domains.

This paper presents trends and performance of opinion retrieval techniques proposed within the last 8 years. We identify major techniques in opinion retrieval and group them into four popular categories. We describe the state-of-the-art techniques for each category and emphasize on their performance and limitations. We then summarize with a performance comparison table for the techniques on different datasets. Finally, we highlight possible future research directions that can help solve existing challenges in opinion retrieval.

This paper discusses a goal-based approach for the engineering of capacity-driven Web services. In this approach, goals are established to first, define the roles that these Web services will play in implementing business applications, second, frame the requirements that will be put on these Web services, and third, identify the processes in terms of business logics that these Web services will carry out. Because of the nature of capacity-driven Web services compared with regular (i.e. mono-capacity) Web services, their engineering in terms of design, development, and deployment takes place in a different way. A Web service that is empowered with several capacities, which are basically separate groups of operations to execute, has to choose one capacity for triggering at run-time. To this end, the Web service takes into account different types of requirements like data and privacy that are put on each capacity that empowers this Web service.

For successful application of knowledge-based systems, “knowledge transfer” (KT) should be receiving more attention. Knowledge transfer is defined as a flow of knowledge from knowledge suppliers through knowledge-based systems to users. Insufficient state-of-the-application of knowledge-based systems in Japan and its near future directions are analysed in terms of KT. Some example tasks for improving KT are reported. These tasks were experienced by the author in the course of 10 years of research and development on knowledge-based systems for project management.

Knowledge collection systems often assume they are cooperating with an unbiased expert. They have few functions for checking and fixing the realism of the expertise transferred to the knowledge base, plan, document or other product of the interaction. The same problem arises when human knowledge engineers interview experts. The knowledge engineer may suffer from the same biases as the domain expert. Such biases remain in the knowledge base and cause difficulties for years to come.

To prevent such difficulties, this paper introduces the reader to “critic engineering”, a methodology that is useful when it is necessary to doubt, trap and repair expert judgment during a knowledge collection process. With the use of this method, the human expert and knowledge-based critic form a cooperative system. Neither agent alone can complete the task as well as the two together.

The methodology suggested here offers a number of extensions to traditional knowledge engineering techniques. Traditional knowledge engineering often answers the questions delineated in generic task (GT) theory, yet GT theory fails to provide four additional sets of questions that one must answer to engineer a knowledge base, plan, design or diagnosis when the expert is prone to error. This extended methodology is called “critic engineering”.

Computer induction is discussed in the light of concerns recently expressed by Bloomfield about the use of such algorithms in expert systems construction.

This is the second of two special issues focusing on the integration of artificial intelligence (AI) and operations research (OR) techniques for solving hard computational problems, with an emphasis on planning and scheduling. Both the AI and the OR community have developed sophisticated techniques to tackle such challenging problems. OR has relied heavily on mathematical programming formulations such as integer and linear programming, while AI has developed constraint-based search techniques and inference methods. Recently, we have seen a convergence of ideas, drawing on the individual strengths of these paradigms.

Negotiation in a multiagent system is a topic of active interest for enabling the allocation of scarce resources among autonomous agents. This paper presents a discussion of the research on negotiation criteria, which puts in context the contributions to resource allocation from the fields of economics, mathematics, and multiagent systems. We group the criteria based on how they relate to each other as well as their historical origin. In addition, we present three new criteria: verifiability, dimensionality, and topology. The criteria are organized into five categories. The allocation category contains criteria concerning fairness and envy-freeness with respect to how resources are allocated to agents. The protocol category covers criteria for stability, strategy-proofness, and communication costs. The procedure category includes criteria about the complexity of allocation procedures. The resource category has criteria for the properties that various resources can take and how they affect allocation. The paper concludes with a discussion of the criteria for agent utility functions. The overall objectives of this paper are (1) to create a starting point for protocol engineering by future negotiation designers and (2) to enumerate the criteria and their measures that enable negotiation and allocation mechanisms to be compared objectively.

A major task of organic chemistry is the synthesis of compounds of interest (“targets”) from available starting compounds. We address here the generation by computer of the best synthesis routes, or sequences of reactions, to such compounds. The number of possible routes in the search space is enormous and so the reasoning is examined in some detail, leading to a protocol for the operation of our program, SYNGEN, which has been under development for some years. It requires a new representation of the data – molecular structures and their reactions – which is digital in format and quite unlike the usual graphical depictions of organic chemistry. This is presented as a problem in artificial intelligence, and every effort has been made to present it with a minimum background of organic chemistry required of the reader.

In this paper, we present an approach to solve the drawbacks of manual composition of software components. Our approach is applied within the jcolibri framework for building case-based reasoning (CBR) applications. We propose a system design process based on reusing templates obtained from previously designed CBR systems. Templates store the control flow of the CBR applications and include semantic annotations conceptualizing its behavior and expertise. We use CBR ontology to formalize syntactical, semantical and pragmatical aspects of the reusable components of the framework. The ontology vocabulary facilitates an annotation process of the components and allows to reason about their composition, facilitating the semi-automatic configuration of complex systems from their composing pieces.

Automated planning is the field of Artificial Intelligence (AI) that focuses on identifying sequences of actions allowing to reach a goal state from a given initial state. The need of using such techniques in real-world applications has brought popular languages for expressing automated planning problems to provide direct support for continuous and discrete state variables, along with changes that can be either instantaneous or durative. PDDL+ (Planning Domain Definition Language +) models support the encoding of such representations, but the resulting planning problems are notoriously difficult for AI planners to cope with due to non-linear dependencies arising from the variables and infinite search spaces. This difficulty is exacerbated by the potentially huge fully ground representations used by modern planners in order to effectively explore the search space, which can make some problems impossible to tackle.

This paper investigates two grounding techniques for PDDL+ problems, both aimed at reducing the size of the full ground representation by reasoning over the lifted, more abstract problem structure. The first method extends the simple mechanism of invariant analysis to limit the groundings of operators upfront. The second method proposes to tackle the grounding process through a PDDL+ to classical planning abstraction; this allows us to leverage the amount of research done in the classical planning area. Our empirical analysis studies the effect of these novel approaches over both real-world hybrid applications and synthetic PDDL+ problems took from standard benchmarks of the planning community; our results reveal that not only the techniques improve the running time of previous grounding mechanisms but also let the planner extend the reach to problems that were not solvable before.

The purpose of this paper is to review the use of knowledge-based systems and artificial intelligence (AI) in business. Part I of this paper provided a broad survey of the use of AI in business, summarizing the application of AI in a number of business domains. In addition, it also provided a summary of the use of different forms of knowledge representation in business applications. Part I has a large set of references, including a number of survey papers, focusing on AI in business. Part II of this paper consists of more detailed analysis of particular systems or issues affecting AI in business. It examines technical issues which are central to the construction of business AI systems, and it also examines the commercial contribution made by methods for the development of AI systems. In addition, part II looks at integration between AI and more traditional information systems. AI can be used to add value to many existing information systems, such as database management systems. Particular attention is given to the integration of AI with operations research, which is the one of the primary “competitors” of AI, providing an alternative set of support tools for decision making.

Business organizations are not concerned only with technology issues; there is also concern about the impact of AI on organizations. Further, the evaluation of AI often is based on an economic view of the world. Part II therefore investigates the organizational impact of AI, and the economics of AI, including issues such as value creation. The format of Part II is as follows: Section 8 analyses techniques for improving the performance of AI systems, thus maximizing economic return. Section 9 looks at different forms of uncertainty and ambiguity which must be dealt with by AI systems. It examines the contributions of fuzzy logic and numerical measures of certainty to handling these problems. Section 10 examines the usefulness of different approaches to knowledge acquisition in business situations, and investigates the benefits of methodological approaches to AI applications. It also looks at more recent AI programming techniques which eliminate the need for knowledge elicitation from an expert: neural networks, case-based reasoning and genetic algorithms are discussed. Sections 11 and 12 examine issues of integrating AI systems. Generally, the use of AI in business settings must ultimately be integrated with the broader base of corporate information systems. Section 11 looks at integration with information systems in general, and section 12 looks particularly at integration with operations research. Sections 13 and 14 review the organizational and economic impact of AI. Finally, section 15 provides a brief summary of part II.

We propose a fully distributed actor-critic architecture, named diffusion-distributed-actor-critic Diff-DAC, with application to multitask reinforcement learning (MRL). During the learning process, agents communicate their value and policy parameters to their neighbours, diffusing the information across a network of agents with no need for a central station. Each agent can only access data from its local task, but aims to learn a common policy that performs well for the whole set of tasks. The architecture is scalable, since the computational and communication cost per agent depends on the number of neighbours rather than the overall number of agents. We derive Diff-DAC from duality theory and provide novel insights into the actor-critic framework, showing that it is actually an instance of the dual-ascent method. We prove almost sure convergence of Diff-DAC to a common policy under general assumptions that hold even for deep neural network approximations. For more restrictive assumptions, we also prove that this common policy is a stationary point of an approximation of the original problem. Numerical results on multitask extensions of common continuous control benchmarks demonstrate that Diff-DAC stabilises learning and has a regularising effect that induces higher performance and better generalisation properties than previous architectures.

The oft quoted inadequacy of von Neumann architectures for AI applications has regularly been used to justify the design of special purpose parallel machines. In particular, the von Neumann computational model has been criticized as being unsuitable for parallelism because of the memory access bottleneck. For the design of a new machine both top-down and bottom-up methodologies have drawbacks. The middle-out strategy, working both up and down from an intrinsically concurrent high-level programming language as a means of both representing and processing knowledge provides an attractive way of providing a symbiosis between software and hardware. The longest established and most well-founded symbolic method for the representation and manipulation of knowledge is logic. A notable result of the last decade, work on mechanical theorem proving was that a subset of predicate logic, Horn Clauses, can form the foundation of a computational model. The execution model of Prolog, the first popular language based on Horn Clauses, was designed for efficient evaluation on von Neumann architectures. An alternative computational model, more suitable for expressing reactive systems but retaining Prolog's affinity for metaprogramming, has given rise to a new class of languages, concurrent logic languages. One among many of these languages, FGHC (flat Guarded Horn Clauses), was developed by the Japanese as the kernel of their Fifth Generation initiative. A background familiarity with Prolog would be helpful in understanding this article.

When a mechanic troubleshoots a car, a physician narrows down a diagnosis, a designer selects from among options, or a corporation conducts a post-mortem of a venture gone wrong, the nature knowledge used is more likely than not, the same. That knowledge is “function”, though in the above scenarios it may be variously referred to as “responsibility”, “causal factor” or “missing element”. Reasoning about function is an integral part of human reasoning. It is currently attracting a lot of attention in artificial intelligence.