Catálogo de publicaciones - libros

Information systems have become the backbone of all kinds of organizations today. In almost every sector—manufacturing, education, health care, government, and businesses—large and small information systems are relied upon for everyday work, communication, information gathering, and decision-making. Yet, the inflexibilities in current technologies and methods have also resulted in poor performance, incompatibilities, and obstacles to change. As many organizations are reinventing themselves to meet the challenges of global competition and e-commerce, there is increasing pressure to develop and deploy new technologies that are flexible, robust, and responsive to rapid and unexpected change.

Data integration is the process by which data from heterogeneous data sources are conceptually integrated into a single cohesive data set. In recent years agents have been increasingly used in information systems to promote performance. In this work we propose a modeling framework for agent oriented data integration to demonstrate how agents can support this process. We provide a systematic analysis of the process using real world scenarios, taken from email messages from citizens in a local government, and demonstrate two agent oriented data integration tasks, email routing and opinion analysis.

Organic computing is an upcoming research area with strong relationships to the ideas and concepts of agent-based systems. In this paper, we therefore will have a closer look at agent systems, organic computing systems (as well as autonomic computing systems) and state commonalities and divergences between them. We then propose a common view on these technologies and show, how they can benefit from each other with regard to software engineering.

One way to make engineering design effective and efficient is to make its processes flexible – i.e. self-adjusting, self-configuring, and self-optimizing at run time. This paper presents the descriptive part of the Dynamic Engineering Design Process (DEDP) modeling framework developed in the PSI project. The project aims to build a software tool to assist managers to analyze and enhance the productivity of the DEDPs through process simulations. The framework incorporates the models of teams and actors, tasks and activities as well as design artifacts as the major interrelated parts. DEDPs are modeled as weakly defined flows of tasks and atomic activities which may only “become apparent” at run time because of several presented dynamic factors. The processes are self-formed through the mechanisms of collaboration in the dynamic team of actors. These mechanisms are based on several types of contracting negotiations. DEDP productivity is assessed by the Units of Welfare collected by the multi-agent system which models the design team. The models of the framework are formalized in the family of DEDP ontologies.

The influence of ontologies in Knowledge Based Systems (KBS) methodologies extends well beyond the initial analysis phase, leading in the 1990s to domain-independent KBS methodologies. In this paper, we reflect on those lessons and on the roles of ontologies in KBS development. We analyse and identify which of those roles can be transferred towards an ontology-based MAS development methodology. We identify ontology-related inter-dependencies between the analysis and design phases. We produce a set of six recommendations towards creating a domain-independent MAS methodology that incorporates ontologies beyond its initial analysis phase. We identify its essential features and sketch the characteristic tasks within both its analysis and design phases.

Multi-agent Domain Engineering is a process for the construction of domain-specific agent-oriented reusable software artifacts, like domain models representing the requirements of a family of multi-agent systems, and frameworks, implementing an agent-oriented solution to those requirements. This work describes DDEMAS, an ontology-based technique for the architectural and detailed design of multi-agent frameworks providing a solution to the requirements of a family of multi-agent software systems specified in a domain model. DDEMAS is part of MADEM, a methodology for domain analysis and design of a family of multi-agent systems in a domain. Domain models and multi-agent frameworks are part of a knowledge base constructed through the instantiation of ONTOMADEM, an ontology that represents the knowledge of MADEM. Some examples from a case study on the application of DDEMAS on the construction of a multi-agent framework for the development of usage mining-based Web personalization systems are also described.

This paper proposes an agent-oriented meta-model that provides rigorous concepts for conducting enterprise modelling. The aim is to allow analysts to produce an enterprise model that precisely captures the knowledge of an organization and of its business processes so that an agent-oriented requirements specification of the system-to-be and its operational corporate environment can be derived from it. To this end, the model identifies constructs that enable capturing the intrinsic characteristics of an agent system such as autonomy, intentionality, sociality, identity and boundary, or rational self-interest; an agent being an organizational actor and/or a software component. Such an approach of the concept of agent allows the analyst to have a holistic perspective integrating human and organizational aspects to gain better understanding of business system inner and outer modelling issues. The meta-model takes roots in both management theory and requirements engineering. It helps bridging the gap between enterprise and requirements models proposing an integrated framework, comprehensive and expressive to both managers and software (requirements) engineers.

Current speech-act based ACLs specify domain-independent information about communication and relegate domain-dependent information to an unspecified content language. This is reasonable, but the ACLs cover only a small fraction of the domain-independent information possible. As a key element of modern ACLs, the set of communicative acts needs to be as complete as possible to enable agents to communicate the widest range of information with agreed-upon semantics. This paper describes a new approach to broaden the semantic coverage of ACL speech acts. It provides agents with the ability to express more of the possible meanings in human languages and yields a more powerful ACL. Specifically, we first compare Austin’s and Searle’s classifications, and Ballmer and Brennenstuhl’s comprehensive classification of speech acts. The main meaning categories and their semantics are given next. Finally, a multifaceted evaluation of our approach is presented, which points out that the approach potentially can combine the benefits of the FIPA ACL with Ballmer and Brennenstuhl’s speech act classification, resulting in a more expressive ACL.

The ability to communicate is one of the crucial properties of agents. In this paper a normative approach to the pragmatics of Agent Communication Languages (ACLs) is proposed. In an open environment, like the Internet, in which agents are designed in many different ways, it is important to clearly establish the meaning of a standard language for artificial agents. Traditionally, the pragmatics of ACLs take the form of interaction protocols, which only specify the order in which messages occur without taking into account the content of the message, or the role of the agents. We present a unified ACL which includes the semantics and pragmatics of ACLs, focusing on a pragmatic level based on the social and normative notion of right. The framework is developed extending CTL with modal and deontic operators, and the pragmatics are expressed by means of a prolog-like declarative language.

As utility calculus cannot account for an important part of agents’ behaviour in Multi-Agent Systems, researchers have progressively adopted a more normative approach. Unfortunately, social laws have turned out to be too restrictive in real-life domains where autonomous agents’ activity cannot be completely specified in advance. The idea of Rights is a halfway concept between anarchic and off-line constrained interaction. Rights improve coordination and facilitate social action in Multi-Agent domains, they allow agents enough freedom, and at the same time constrain them (prohibiting specific actions). So far rights have not been tested or proven experimentally. We are comparing experimentally the three mentioned interaction architectures in the domain of agent-based traffic simulation.