Content uploaded by Enrique Paniagua Arís
Author content
All content in this area was uploaded by Enrique Paniagua Arís on Jan 28, 2016
Content may be subject to copyright.
A preview of the PDF is not available
Contribución a la representación y generación de planes con incertidumbre
Abstract
Planificación de Desarrollo Cooperativo (PDC) es un modelo de planificación de Proyectos de Cooperación y Desarrollo, implementado en el ONGIA, una prototipo de Inteligencia Artificial para el estudio predictive de Proyectos de Desarrollo. Los Proyectos de Cooperación y Desarrollo realizados por las ONGDs (Organizaciones No Gubernamentales de Desarrollo) están marcados por un alto grado de incertidumbre, debido por una parte, a la multiplicidad de intereses de los diferentes agentes que intervienen en la definición y ejecución del proyecto, así como a la naturaleza dinámica del contexto (entorno) que convive con la evolución del proyecto, por la otra. Es por este motivo que se propone el modelo de Planificación de Desarrollo Cooperativo, para la simulación de las fases: Identificación de los Problemas, Definición de los Objetivos, Generación de los Planes Alternativos y Viabilidad de los Planes, que nos permitirá consensuar a los distintos agentes que intervienen a lo largo del desarrollo de un Proyecto de Cooperación y Desarrollo. ONGIA está basado en una arquitectura de agentes distribuidos. Cada agente dispone de un conocimiento especializado (el específico de cada grupo de usuarios del proyecto); una lógica común a todo el grupo de agentes (un álgebra de valores de verdad que parte de los conjuntos borrosos); y un conjunto de metareglas de control. El mecanismo de consenso en la definición del problema y objetivos en el ONGIA se basa en la teoría de los Sistemas de Argumentación y la de los Sistemas Multicontexto. El modelo para consensuar a los diferentes agentes en la definición de los planes abstractos alternativos utiliza la función de Beneficio Conjunto para llegar al compromiso entre los distintos ejecutores del plan abstracto. La técnica utilizada para controlar el problema de la cualifícación se basa en una Ordenación Aproximada Priorizada. Finalmente, el modelo para simular la ejecución de cada plan abstracto se basa en la teoría de las Situaciones Posibles y en la de la Entropía Termodinámica para describir la posible evolución del plan.
En el presente articulo se presenta el analisis de los Sistemas MultiAgente (SMA de ahora en adelante) como alternativa para el modelado de la actuacion en organizaciones humanas. Para poder realizar dicho analisis, primero debemos de tener en cuenta los componentes de conocimiento requeridos en una organizacion humana, para despues, especificar las distintas caracteristicas que debe poseer un SMA en el modelado de la actuacion de dicha organizacion, principalmente en el componente del Capital Humano; posteriormente, debemos analizar y comparar los distintos modelos de SMA frente a cada una de las caracteristicas descritas previamente; en tercer lugar, debemos analizar la existencia de metodologias para el analisis y diseño de los SMA acordes con los requerimientos propuestos; en cuarto lugar debemos observar el estado de la tecnologia de cara a poder implementar dichas actitudes del SMA. Y finalmente, como efecto lateral, nos podemos preguntar que caracteristicas (o Modelos de Gestion Empresarial) tienen las organizaciones humanas de las que podamos proponer nuevas lineas en el desarrollo de los SMA.
This paper presents a new category of social agents in multiple
agent systems. This new class of agent, named teleo-social agents, is
concerned with both maintenance and protection of the group central
goal. The knowledge level of the teleo-social agent consists of a) the
components, composition laws, behaviour laws and medium of the group, b)
the social tasks the agent can apply to it, and c) a social
balance-oriented control model. This type of agent is introduced by
referring to its use in simulation of development projects
The Maximum Entropy Method and the related Bayesian Probability Theory have recently become somewhat better known to research workers; but the power of these methods to resolve the true structures underlying many kinds of data is still not as widely recognized as it ought to be. This book is the first readily available account of such methods. It grew out of a series of interdisciplinary seminars, organized by the editors, and given by leading proponents of maximum entropy techniques in the UK. The written versions are intended to introduce the reader to the principles of probability theory, viewed as the logic of inference, and to explain the role that entropy plays in the assignment of probabilities. The chapters concerned with applications show how the resulting algorithms can be implemented and how they work out in practice. In addition, new light is shed upon the relation between statistical and dynamical theories in physics, and on the confusion which surrounds the Second Law of Thermodynamics.
The frame problem is about the things that stay the same and not the things that change. The two are not equivalent. The frame problem is more general than the situation calculus. It seems to appear in just about any temporal notation. When it does appear, it puts a severe strain on the inferential resources available in a deductive framework, even when deduction is extended non-monotonically. On the other hand, it is not obvious that the frame problem is a serious issue outside the context of logic. This chapter discusses inertia problem and ramification problem. The ramification problem is dealt with by selecting a model of the current situation and using it to make inferences. The inertia problem has always looked like an excellent opportunity to apply non-monotonic logic which, as everyone knows by now, is the name given to logical systems in which adding axioms can cause proofs to disappear. Most AI practitioners confront and solve the inertia problem without even noticing it. It would be different if theorem proving played a large role in problem solving, but it does not. The inertia problem is of interest mainly to a fringe group, those who believe that logical analyses are relevant to building knowledge representations. Logic is valuable as a tool for organizing the semantics of knowledge representations. If the inertia problem needs to be solved, then model theory may well prove indispensable to its solution.
Humans and intelligent computer programs must often jump to the conclusion that the objects they can determine to have certain properties or relations are the only objects that do. Circumscription formalizes such conjectural reasoning.
Artificial Intelligence is an interdisciplinary field, and formed about 60 years ago as an interaction between mathematical methods, computer science, psychology, and linguistics. Artificial Intelligence is an experimental science and today features a number of internally designed theoretical methods: knowledge representation, modeling of reasoning and behavior, textual analysis, and data mining. Within the framework of Artificial Intelligence, novel scientific domains have arisen: non-monotonic logic, description logic, heuristic programming, expert systems, and knowledge-based software engineering. Increasing interest in Artificial Intelligence in recent years is related to the development of promising new technologies based on specific methods like knowledge discovery (or machine learning), natural language processing, autonomous unmanned intelligent systems, and hybrid human-machine intelligence.
One objective of distributed artificial intelligence research is to build systems that are capable of cooperative problem solving. To this end, a number of implementation-oriented models of cooperative problem solving have been developed. However, mathematical models of social activity have focussed only on limited aspects of the cooperative problem solving process: no mathematical model of the entire process has yet been described. In this paper, we rectify this omission. We present a preliminary model that describes the cooperative problem solving process from recognition of the potential for cooperation through to team action. The model is formalised by representing it as a theory in a quantified multimodal logic. A key feature of the model is its reliance on the twin notions of commitments and conventions; conventions (protocols for monitoring commitments) are formalised for the first time in this paper. We comment on the generality of the model, outline its deficiencies, and suggest some possible refinements and other future areas of research.
This paper proposes, characterizes and outlines the benefits of a new computer level specifically for multi-agent problem solvers. This level is called the cooperation knowledge level and involves describing and developing richer and more explicit models of common social phenomena. We then focus on one particular form of social interaction in which groups of agents decide they wish to work together, in a collaborative manner, to tackle a common problem. A domain independent model (called joint responsibility) is developed to describe how participants should behave during such problem solving. Particular emphasis is placed on the problem of ensuring coherent behaviour in the face of unpredictable and dynamic environments. The utility of this model is highlighted in the real-world environment of electricity transportation management. In this domain, agents have to make decisions using partial, imprecise views of the system and the inherent dynamics of the problem mean that team members have to continually evaluate the ongoing problem solving process. Joint responsibility provides the evaluation criteria and the causal link to behaviour upon which individual and social situation assessment is based.
The concept of an agent has recently become important in Artificial Intelligence (AI), and its relatively youthful subfield, DistributedAI (DAI). Our aim in this paper is to point the reader at what we perceive to be the most important theoretical and practical issues associated with the design and construction of intelligent agents. For convenience, we divide the area into three themes (though as the reader will see, these divisions are at times somewhat arbitrary). Agent theory is concerned with the question of what an agent is, and the use of mathematical formalisms for representing and reasoning about the properties of agents. Agent architectures can be thought of as software engineering models of agents; researchers in this area are primarily concerned with the problem of constructing software or hardware systems that will satisfy the properties specified by agent theorists. Finally, agent languages are software systems for programming and experimenting with agents; these languages typically embody principles proposed by theorists. The paper is not intended to serve as a tutorial introduction to all the issues mentioned; we hope instead simply to identify the key issues, and point to work that elaborates on them. The paper closes with a detailed bibliography, and some biblographical remarks.
One objective of distributed artificial intelligence research is to build systems that are capable of cooperative problem solving. To this end, a number of implementation oriented models of cooperative problem solving have been developed. However, mathematical models of social activity have focussed only on limited aspects of the cooperative problem solving process: no mathematical model of the entire process has yet been described. In this paper, we rectify this omission. We present a preliminary model that describes the cooperative problem solving process from recognition of the potential for cooperation through to team action. The model is formalised by representing it as a theory in a quantified multi-modal logic. A key feature of the model is its reliance on the twin notions of commitments and conventions; conventions (protocols for monitoring commitments) are formalised for the first time in this paper. We comment on the generality of the model, outline its deficiencies, and suggest some possible refinements and other future areas of research.