The first International and Interdisciplinary Conference on Modeling and Using Context (CONTEXT-97) was held at Rio de Janeiro, Brazil on February 4–6 1997. This article provides a summary of the presentations and discussions during the three days with a focus on context in applications. The notion of context is far from defined, and is dependent in its interpretation on a cognitive science versus an engineering (or system building) point of view. However, the conference makes it possible to identify new trends in the formalization of context at a theoretical level, as well as in the use of context in real-world applications. Results presented at the conference are ascribed in the realm of the works on context over the past few years at specific workshops and symposia. The diversity of the attendees' origins (artificial intelligence, linguistics, philosophy, psychology, etc.) demonstrates that there are different types of context, not a unique one. For instance, logicians model context at the level of the knowledge representation and the reasoning mechanisms, while cognitive scientists consider context at the level of the interaction between two agents (i.e. two humans or a human and a machine). In the latter case, there are now strong arguments proving that one can speak of context only in reference to its use (e.g. context of an item or of a problem solving exercise). Moreover, there are different types of context that are interdependent. This makes it possible to understand why, despite the consensus on some context aspects, agreement on the notion of context is not yet achieved.

A challenging issue to advance the existing P2P semantic routing protocols is related to the capability of developing mechanisms for focused selection of the query recipients by taking into account a semantically rich description of the context of each peer. In this article, we present the H-Link semantic routing approach designed to exploit the results of an ontology matchmaking process for providing a semantic overlay network where peers having similar contexts are recognized and interlinked as semantic neighbors. In particular, H-Link aims at advancing the existing semantic routing protocols by combining ontology-based peer context descriptions and ontology matching techniques for providing query forwarding on a real semantic basis, in a completely decentralized way.

The claim is frequently made that human judgement and reasoning are vulnerable to cognitive biases. Such biases are assumed to be inherent in that they are attributed to the nature of the mental processes that produce judgement. In this paper, we review the psychological evidence for this claim in the context of the debate concerning human judgemental competence under uncertainty. We consider recent counter-arguments which suggest that the evidence for cognitive biases may be dependent on observations of performance on inappropriate tasks and by comparisons with inappropriate normative standards. We also consider the practical implications for the design of decision support systems.

The use of agents in electronic commerce has been explored greatly over the past several years. A large majority of this effort is toward commerce where businesses have direct transactions with consumers (B2C). However, the transactions that occur between businesses (B2B) are far more prevalent than B2C. Research where agents are used for B2B can be classified in five basic areas, service discovery, mediation, negotiation, process management (be it workflow or supply-chain management), and evaluation. At the 2001 International Bi-Conference Sessions on Agent-Based Approaches to B2B Interoperability (AgentB2B), practitioners were invited to present their research and industry efforts in each of these areas. This paper summaries the work and conclusions presented at these two events.

Recent theoretical results have justified the use of potential-based reward shaping as a way to improve the performance of multi-agent reinforcement learning (MARL). However, the question remains of how to generate a useful potential function.

Previous research demonstrated the use of STRIPS operator knowledge to automatically generate a potential function for single-agent reinforcement learning. Following up on this work, we investigate the use of STRIPS planning knowledge in the context of MARL.

Our results show that a potential function based on joint or individual plan knowledge can significantly improve MARL performance compared with no shaping. In addition, we investigate the limitations of individual plan knowledge as a source of reward shaping in cases where the combination of individual agent plans causes conflict.

We have produced an ontology specifying a model of computer attack. Our ontology is based upon an analysis of over 4000 classes of computer intrusions and their corresponding attack strategies and is categorised according to system component targeted, means of attack, consequence of attack and location of attacker. We argue that any taxonomic characteristics used to define a computer attack be limited in scope to those features that are observable and measurable at the target of the attack. We present our model as a target-centric ontology that is to be refined and expanded over time. We state the benefits of forgoing dependence upon taxonomies in favour of ontologies for the classification of computer attacks and intrusions. We have specified our ontology using the DARPA Agent Markup Language+Ontology Inference Layer and have prototyped it using DAMLJessKB. We present our model as a target-centric ontology and illustrate the benefits of utilising an ontology in lieu of a taxonomy, by presenting a use-case scenario of a distributed intrusion detection system.

This paper presents a personal interpretation of the evolution of artificial intelligence (AI) systems during these last 25 years. This evolution is presented along five generations of AI systems, namely expert systems, joint cognitive systems, intelligent systems, intelligent assistant systems, and the coming generation of context-based intelligent assistant systems. Our testimony relies on different real-world applications in different domains, especially for the French national power company, the subway companies in Paris and in Rio de Janeiro, in medicine, a platform for e-maintenance, road safety, and open sources. Our main claim is to underline that the next generation of AI systems (context-based intelligent assistant systems) requires a radically different consideration on context and its relations with the users, the task at hand, the situation, and the environment in which the task is accomplish by the user; the observation of users through their behaviors and not a profile library; a robust conceptual framework for modeling and managing context; and a computational tool for representing in a uniform way pieces of knowledge, of reasoning, and of contexts.

Biologists depend on visual representations, and their use of diagrams has drawn the attention of philosophers, historians, and sociologists interested in understanding how these images are involved in biological reasoning. These studies, however, proceed from identification of diagrams on the basis of their spare visual appearance, and do not draw on a foundational theory of the nature of diagrams as representations. This approach has limited the extent to which we understand how these diagrams are involved in biological reasoning. In this paper, I characterize three different kinds of figures among those previously identified as diagrams. The features that make these figures distinctive as representational types, furthermore, illuminate the ways in which they are involved in biological reasoning.

In spite of the rapid spread of agent technology, there is, as yet, little evidence of an engineering approach to the development of agent-based systems. In particular, development methods for these systems are relatively rare. One of the key reasons for this is the inadequacy of standard software development approaches for these new, and fundamentally different, agent-based systems. Traditional software development methods often lack the flexibility to handle high-level concepts such as an agent's dynamic control of its own behaviour, its ability to represent cooperative interactions, and its mechanisms for representing internal change, assumptions, objectives, and the uncertainty inherent in its interactions with the real-world.

Reinforcement learning (RL) can be extremely effective in solving complex, real-world problems. However, injecting human knowledge into an RL agent may require extensive effort and expertise on the human designer’s part. To date, human factors are generally not considered in the development and evaluation of possible RL approaches. In this article, we set out to investigate how different methods for injecting human knowledge are applied, in practice, by human designers of varying levels of knowledge and skill. We perform the first empirical evaluation of several methods, including a newly proposed method named State Action Similarity Solutions (SASS) which is based on the notion of similarities in the agent’s state–action space. Through this human study, consisting of 51 human participants, we shed new light on the human factors that play a key role in RL. We find that the classical reward shaping technique seems to be the most natural method for most designers, both expert and non-expert, to speed up RL. However, we further find that our proposed method SASS can be effectively and efficiently combined with reward shaping, and provides a beneficial alternative to using only a single-speedup method with minimal human designer effort overhead.

In this paper, we discuss the potential role of answer set programming (ASP) in the context of approaches to the development of agents and multi-agent systems especially in the realm of Computational Logic. After shortly recalling the main (computational-logic-based) agent-oriented frameworks, we introduce ASP; then, we discuss the usefulness of a potential integration of the two paradigms in a modular heterogeneous framework, and the feasibility of such integration. This also in the more general view of improving and empowering flexibility of agent-oriented frameworks. Relevant literature will be mentioned and discussed. Possible future directions and potential developments will be outlined.

Rough set theory (RST) has enjoyed an enormous amount of attention in recent years and has been applied to many real-world problems including data mining, pattern recognition, and intelligent control. Much research has recently been carried out in respect of both the development of the underlying theory and the application to new problem domains. This paper attempts to summarize the advances in RST, its extensions, and their applications. It also identifies important areas which require further investigation. Typical example application domains are examined which demonstrate the success of the application of RST to a wide variety of areas and disciplines, and which also exhibit the strengths and limitations of the respective underlying approaches.

The relentless growth of the computer industry over more than 30 years has been driven on by a series of major innovations. High-level languages; solid-state logic; compatible machine ranges; disk storage; time-sharing; data communications; virtual machine architectures; the use of LSI and solid-state memory; text processing; personal computers; sucessful packaged software; each advance in its turn has opened up new markets and set off a new spurt of expansion. A vital conclusion of this study is that expert systems promise to be another such major innovation.

To foster effective use of artificial intelligence planning and scheduling (P&S) systems in the real world, it is of great importance to both (a) broaden direct access to the technology for the end users and (b) significantly increase their trust in such technology. Automated P&S systems often bring solutions to the users that are neither ‘obvious’ nor immediately acceptable to them. This is because these tools directly reason on causal, temporal, and resource constraints; moreover, they employ resolution processes designed to optimize the solution with respect to non-trivial evaluation functions. Knowledge engineering environments aim at simplifying direct access to the technology for people other than the original system designers, while the integration of validation and verification (V&V) capabilities in such environments may potentially enhance the users’ trust in the technology. Somehow, V&V techniques may represent a complementary technology, with respect to P&S, that contributes to developing richer software environments to synthesize a new generation of robust problem-solving applications. The integration of V&V and P&S techniques in a knowledge engineering environment is the topic of this paper. In particular, it analyzes the use of state-of-the-art V&V technology to support knowledge engineering for a timeline-based planning system called MrSPOCK. The paper presents the application domain for which the automated solver has been developed, introduces the timeline-based planning ideas, and then describes the different possibilities to apply V&V to planning. Hence, it continues by describing the step of adding V&V functionalities around the specialized planner, MrSPOCK. New functionalities have been added to perform both model validation and plan verification. Lastly, a specific section describes the benefits as well as the performance of such functionalities.

Several Apriori algorithm implementations for mining association rules have been proposed in the literature using the Hadoop-MapReduce framework and, more recently, Spark. However, none of the works have made a detailed assessment of its performance, for example, comparing it with other implementations in various characteristics of data sets. In this work, we present a review of the main algorithms proposed for Hadoop-MapReduce and compared their implementations in a single environment under several different situations. Moreover, these algorithms had their implementations adapted to Spark, and also compared under the same circumstances. Based on the results of the experiments, we present a framework for recommending the Apriori implementation most appropriate for solving a given problem, according to the data set characteristics and minimum required support. The results show that Spark implementations overcome Hadoop-MapReduce implementations at runtime in most experiments. However, there is no single implementation that is the best in all the evaluated situations.

The computer industry is in the early stages of a revolution. Its capability for processing natural languages will advance dramatically over the next few years. As a result, computer systems which deal with human language will begin to play a central role in the industry. Computers will becooome significantly more accessible to many people, and many tasks which are wholly manual will be handled electronically.

Planning systems have been an active research topic within Artificial Intelligence for over two decades. There have been a number of techniques developed during that period which still form an essential part of many of today's planners. This paper introduces the techniques, attempts to classify some of the important research themes in AI planning and describes their historical development.