Dynamic Specifications for Norm-Governed Systems

Dynamic Specifications for Norm-Governed Systems

Alexander Artikis (National Centre for Scientific Research “Demokritos”, Greece), Dimosthenis Kaponis (Imperial College London, UK) and Jeremy Pitt (Imperial College London, UK)
DOI: 10.4018/978-1-60566-256-5.ch019
OnDemand PDF Download:
$37.50

Abstract

We have been developing a framework for executable specification of norm-governed multi-agent systems. In this framework, specification is a design-time activity; moreover, there is no support for run-time modification of the specification. Due to environmental, social, or other conditions, however, it is often desirable, or even necessary, to alter the system specification during the system execution. In this chapter we extend our framework by allowing for “dynamic specifications”, that is, specifications that may be modified at run-time by the members of a system. The framework extension is motivated by Brewka’s “dynamic argument systems”—argument systems in which the rules of order may become the topic of the debate. We illustrate our framework for dynamic specifications by presenting: (i) a dynamic specification of an argumentation protocol, and (ii) an execution of this protocol in which the participating agents modify the protocol specification.
Chapter Preview
Top

Introduction

A particular kind of Multi-Agent System (MAS) is one where the member agents are developed by different parties, and where there is no direct access to an agent’s internal state. In this kind of MAS it cannot be assumed that all agents will behave according to the system specification because the agents act on behalf of parties with competing interests, and thus they may inadvertently fail to, or even deliberately choose not to, conform to the system specification in order to achieve their individual goals. A few examples of this type of MAS are Virtual Organisations, electronic marketplaces, argumentation (dispute resolution) protocols, and negotiation protocols. MAS of this type are often classified as ‘open’.

We have been developing executable specifications of open MAS (Artikis, 2003; Artikis, Sergot & Pitt, 2003; 2007); we adopt a bird’s eye view of these systems, as opposed to an agent’s own perspective whereby it reasons about how it should act. Furthermore, we view agent systems as instances of normative systems (Jones & Sergot, 1993). A feature of this type of system is that actuality, what is the case, and ideality, what ought to be the case, do not necessarily coincide. Therefore, it is essential to specify what is permitted, prohibited, and obligatory, and perhaps other more complex normative relations that may exist between the agents. Amongst these relations, we place considerable emphasis on the representation of institutionalised power (Jones & Sergot, 1996) — a standard feature of any norm-governed system whereby designated agents, when acting in specified roles, are empowered by an institution to create specific relations or states of affairs (such as when an agent is empowered by an institution to award a contract and thereby create a bundle of normative relations between the contracting parties). We encode specifications of open MAS in executable action languages from the field of Artificial Intelligence (Giunchiglia, Lee, Lifschitz, McCain & Turner, 2004; Kowalski & Sergot, 1986).

Our executable specifications may be classified as ‘static’, in the sense that there is no support for their run-time modification. In some open MAS, however, environmental, social or other conditions may favour, or even require, specifications modifiable during the system execution. Consider, for instance, the case of a malfunction of a large number of sensors in a sensor network, or the case of manipulation of a voting procedure due to strategic voting, or when an organisation conducts its business in an inefficient manner. Therefore, we present in this chapter an infrastructure for ‘dynamic specifications’, that is, specifications that are developed at design-time but may be modified at run-time by the members of a system. The presented infrastructure is motivated by Brewka’s ‘dynamic argument systems’ (Brewka, 2001) — argument systems in which, at any point in the disputation, participants may start a meta level debate, that is, the rules of order can become the current point of discussion, with the intention of altering these rules.

Our infrastructure for dynamic specifications allows protocol participants to alter the rules of a protocol P during the protocol execution. P is considered an ‘object’ protocol; at any point in time during the execution of the object protocol the participants may start a ‘meta’ protocol in order to decide whether the object protocol rules should be modified: add a new rule-set, delete an existing one, or replace an existing rule-set with a new one. Moreover, the participants of the meta protocol may initiate a meta-meta protocol to decide whether to modify the rules of the meta protocol, or they may initiate a meta-meta-meta protocol to modify the rules of the meta-meta protocol, and so on.

Key Terms in this Chapter

Hierarchical Structure: A pyramid-shaped system that arranges the relations between the entities within an organization in a top-down way. Power, responsibility and authority are concentrated at the top of the pyramid and decisions flow from the top downwards. The pyramid can be more steep or more flat. A steep pyramid has many layers of management, a flat organization has relatively few (Companion to Organizations, J. Baum, 2002).

Organizational Performance: Comprises the actual output or results of an organization as measured against its intended outputs (or goals and objectives). It is a broad construct which captures what organizations do, produce, and accomplish for the various constituencies with which they interact. Specialists in many fields are concerned with organizational performance including strategic planners, operations, finance, legal, and organizational development (Companion to Organizations, J. Baum Eds., Oxford Blackwell, UK, 2002).

Generative Interactions: Interactions are generative to the extent that they allow for the emergence of new capabilities to handle complexity, notably the increased complexity of signals from the environment. They are used to gain additional knowledge and insight. The value of interactions is rising because their generative function has become the solution to increasingly challenging organizational problems that go far beyond coordination needs (Morieux et al, 2005).

Routines: In the economics and business literatures, the notion of organizational routine has come to stand for regularity in economic activity. The concept of organizational routine is used to capture repetitive, stable activity leading to behavior patterns or recurrent interaction patterns. However the term is used also refering to some cognitive representation such as rules and cognitive. “We will regard a set of activities as routinized, [then,] to the degree that choice has been simplified by the development of a fixed response to defined stimuli. If search has been eliminated, but a choice remains in the form of clearly defined and systematic computing routine, we will say that the activities are routinized” (March and Simon 1993, page 142).

Bounded Rationality: A term for the phenomenon that cognitive blinders prevent people from seeing, seeking, using, or sharing relevant, accessible, and perceivable information during decision-making. The bounded rationality phenomenon challenges traditional rationalist perspectives and suggests that the rationality of actual human and company behavior is always partial, or ‘bounded’ by human limitations. This concept recognizes that decision making takes place within an environment of incomplete information and uncertainty. Herbert Simon pointed out that most people are only partly rational, and are in fact emotional and irrational in the remaining part of their actions. They experience limits in formulating and solving complex problems and in processing (receiving, storing, retrieving, transmitting) information (Companion to Organizations, J. Baum Eds., Oxford Blackwell, UK, 2002).

Span of Control: Refers to how relationships are structured between leaders and subordinates in organizations. It represents the number of people/subordinates that can be effectively managed by one manager. The optimal Span of Control is dependent upon the nature of the work of the subordinates, the skills, capabilities, experience, seniority, qualifications of the managers and subordinates, the use of information technology, the detail at which work rules and procedures have been formalized and are known by the subordinates, the applied management style and the desired depth of the hierarchy in an organization (Companion to Organizations, J. Baum Eds., Oxford Blackwell, UK, 2002).

Satisficing: In economics, satisficing is a behavior which attempts to achieve at least some minimum level of a particular variable, but which does not necessarily maximize its value. The most common application of the concept in economics is in the behavioural theory of the firm, which, unlike traditional accounts, postulates that producers treat profit not as a goal to be maximized, but as a constraint. Under these theories, a critical level of profit must be achieved by firms; thereafter, priority is attached to the attainment of other goals. The word satisfice was coined by Herbert Simon as a portmanteau of “satisfy” and “suffice”. Simon pointed out that human beings lack the cognitive resources to maximize: we usually do not know the relevant probabilities of outcomes, we can rarely evaluate all outcomes with sufficient precision, and our memories are weak and unreliable.

Allocative Interactions: Are used to coordinate events, functions, businesses, etc. such that these fit together and fit with a pre-existing scheme while minimizing the time, energy, etc. consumed to ensure the fit (Morieux et al, 2005).

Near Decomposability: According to H. Simon basically all viable systems, be they physical, social, biological, artificial, share the property of having a near decomposable architecture: they are organized into hierarchical layers of parts, parts of parts, parts of parts of parts and so on, in such a way that interactions among elements belonging to the same parts are much more than interactions among elements belonging to different parts. By “intense” interaction is meant that the behavior of one component depends more closely on the behavior of other components belonging to the same part than on components belonging to other parts (i.e. the cross-derivatives are larger within a part). This kind of architecture can be found in business firms, where division of labor, divisionalization, hierarchical decomposition of tasks are all elements which define a near decomposable system: individuals within a hierarchical subunit have closer, more widespread, more intense and more frequent interactions than individuals belonging to different subunits. But a very similar architecture can also be found in most complex artifacts (which are made by assembling parts and components, which in turn can be assemblies of other parts and components, and so on), in software (with the use of subroutines, and even more so in object-oriented programming) (Egidi and Marengo, 2006).

Complete Chapter List

Search this Book:
Reset