Academia.eduAcademia.edu

Figure 5 – uploaded by Tina Balke-Visser

See full PDFdownloadDownload figure
fable 2: Comparison of Artificial Institution Models - Part II

Table 5 fable 2: Comparison of Artificial Institution Models - Part II

Related Figures (18)

Figure 2: Savarimuthu and Cranefield’s Norm Life Cycle Model 2.2.2. Norm Creation Unlike Finnemore and Sikkink [1998], who acknowledged only the role of norm en- trepreneurs for the creation of norms, Savarimuthu and Cranefield [2011] realize that in the context of multi-agent systems, norms can be created by three different approaches: off- line design, norm leaders and norm entrepreneurs. In off-line design the norm is introduced by an external designer and is hard-wired into the agents. Norm leaders, on the other hand, are powerful agents of the system that (following a democratic or an authoritarian process)
Figure 2: Savarimuthu and Cranefield’s Norm Life Cycle Model 2.2.2. Norm Creation Unlike Finnemore and Sikkink [1998], who acknowledged only the role of norm en- trepreneurs for the creation of norms, Savarimuthu and Cranefield [2011] realize that in the context of multi-agent systems, norms can be created by three different approaches: off- line design, norm leaders and norm entrepreneurs. In off-line design the norm is introduced by an external designer and is hard-wired into the agents. Norm leaders, on the other hand, are powerful agents of the system that (following a democratic or an authoritarian process)
of the complete life cycle. Where existing, the superprocesses are represented as boxes com- prising their elementary processes, with the corresponding superprocess label highlighted in bold font. We will briefly outline the entire life cycle before introducing the individual processes in greater detail. Initially, potential norms are explicitly created, before being transmitted to the wider society, and rely on recognition and enforcement processes (captured in the superprocess enforcement) to promote their adoption. The superprocess internalization involves the deci- sion whether to accept a norm, potentially modifying it, and finally, internalizing it, and thus becoming an enforcer of the norm itself. The subsequent cyclic reinforcement of the norm, including transmission, enforcement and internalization (tagged emergence), determines whether the initial potential norm becomes a norm. If attaining normative status, norms undergo a continuous refinement that requires reiteration through the elementary processes 0 gain salience. Any norm modification, such as the adaptation to new circumstances, im- plies that some normative content is forgotten. Swipe-card payments for bus services, for example, make it increasingly permissible for individuals to enter buses through arbitrary doors, instead of requiring the traditional entry through specific doors for payment. Con- rasting the gradual forgetting of normative content, norms can be superseded by alternative norms, in which case the original norm is forgotten in its entirety. For example, over the past decades in many Western countries the general tolerance towards smoking in public places has been progressively replaced with general rejection. In the following, we will discuss selected processes in greater detail and contextualize
of the complete life cycle. Where existing, the superprocesses are represented as boxes com- prising their elementary processes, with the corresponding superprocess label highlighted in bold font. We will briefly outline the entire life cycle before introducing the individual processes in greater detail. Initially, potential norms are explicitly created, before being transmitted to the wider society, and rely on recognition and enforcement processes (captured in the superprocess enforcement) to promote their adoption. The superprocess internalization involves the deci- sion whether to accept a norm, potentially modifying it, and finally, internalizing it, and thus becoming an enforcer of the norm itself. The subsequent cyclic reinforcement of the norm, including transmission, enforcement and internalization (tagged emergence), determines whether the initial potential norm becomes a norm. If attaining normative status, norms undergo a continuous refinement that requires reiteration through the elementary processes 0 gain salience. Any norm modification, such as the adaptation to new circumstances, im- plies that some normative content is forgotten. Swipe-card payments for bus services, for example, make it increasingly permissible for individuals to enter buses through arbitrary doors, instead of requiring the traditional entry through specific doors for payment. Con- rasting the gradual forgetting of normative content, norms can be superseded by alternative norms, in which case the original norm is forgotten in its entirety. For example, over the past decades in many Western countries the general tolerance towards smoking in public places has been progressively replaced with general rejection. In the following, we will discuss selected processes in greater detail and contextualize
Figure 4: Mahmoud et al.’s Norm Life Cycle Model Overview The latest life cycle model has been proposed by Mahmoud et al. [2014b]. Similar to the earlier life cycle models developed in the context of NorMAS, their work is based on a comprehensive literature review, both considering individual works as well as previous life cycle models. In contrast to Hollander and Wu’s detailed model, their approach identifies five core processes (Creation, Emergence, Assimilation, Internalization, Removal) with a further decomposition of selected processes as shown in Figure 4. Since this model has only been briefly described by the original authors themselves and strongly builds on concepts introduced in the context of Hollander and Wu’s earlier, more detailed model, we provide a concise overview at this stage, before discussing the novel contributions in more detail.
Figure 4: Mahmoud et al.’s Norm Life Cycle Model Overview The latest life cycle model has been proposed by Mahmoud et al. [2014b]. Similar to the earlier life cycle models developed in the context of NorMAS, their work is based on a comprehensive literature review, both considering individual works as well as previous life cycle models. In contrast to Hollander and Wu’s detailed model, their approach identifies five core processes (Creation, Emergence, Assimilation, Internalization, Removal) with a further decomposition of selected processes as shown in Figure 4. Since this model has only been briefly described by the original authors themselves and strongly builds on concepts introduced in the context of Hollander and Wu’s earlier, more detailed model, we provide a concise overview at this stage, before discussing the novel contributions in more detail.
Table 1: Chronological Overview of Literature and Associated Life Cycle Characteristics
Table 1: Chronological Overview of Literature and Associated Life Cycle Characteristics
Figure 5: Schematic Comparison of Discussed Norm Life Cycle Models MODELING NORM DYNAMICS IN MULTI-AGENT SYSTEMS
Figure 5: Schematic Comparison of Discussed Norm Life Cycle Models MODELING NORM DYNAMICS IN MULTI-AGENT SYSTEMS
As stated before, norms can either be explicitly created or identified at runtime (the corresponding right-facing arrows in Figure 6 mark these life cycles starting points). If cre- ated, norms are transmitted and identified.!2 As mentioned above, identification is not only initiated by transmission, but may involve the identification of an existing norm (e.g. by observation). Once internalized (by a complex internalization process that may contain slementary processes as laid out by Hollander and Wu [2011b]), norms can be reinforced. which may operate internally (e.g. based on motivational enforcement or elicited emotions).
As stated before, norms can either be explicitly created or identified at runtime (the corresponding right-facing arrows in Figure 6 mark these life cycles starting points). If cre- ated, norms are transmitted and identified.!2 As mentioned above, identification is not only initiated by transmission, but may involve the identification of an existing norm (e.g. by observation). Once internalized (by a complex internalization process that may contain slementary processes as laid out by Hollander and Wu [2011b]), norms can be reinforced. which may operate internally (e.g. based on motivational enforcement or elicited emotions).
Figure 7: Taxonomy of Norm Synthesis Approaches on approaches that emphasize the detection and identification of norms. notion of norms in the broad sense (i.e. as institutions), ranging from self-enforcing conven- tions via socially enforced norms to centrally enforced social laws or rules. In the following, we will discuss selected contributions to the area of norm synthesis, with a particular focus on approaches that emphasize the detection and identification of norms.
Figure 7: Taxonomy of Norm Synthesis Approaches on approaches that emphasize the detection and identification of norms. notion of norms in the broad sense (i.e. as institutions), ranging from self-enforcing conven- tions via socially enforced norms to centrally enforced social laws or rules. In the following, we will discuss selected contributions to the area of norm synthesis, with a particular focus on approaches that emphasize the detection and identification of norms.
Figure 8: Norm Synthesis Approaches and Related Life Cycle Processes
Figure 8: Norm Synthesis Approaches and Related Life Cycle Processes
Table 1: Comparison of Artificial Institution Models - Part I MODELING ORGANIZATIONS AND INSTITUTIONS IN MAS
Table 1: Comparison of Artificial Institution Models - Part I MODELING ORGANIZATIONS AND INSTITUTIONS IN MAS
Table 3: Comparison of Artificial Organization Models
Table 3: Comparison of Artificial Organization Models
MODELING NORMS EMBEDDED IN SOCIETY
MODELING NORMS EMBEDDED IN SOCIETY
ffirmative. The message exchange process consists of three elements first a messag pace M, second a group equilibrium message correspondence p, denoted uw: Or A nd third outcome function h, denoted ash: M+> Z. A message space M consist f the messages available for communication. Messages may include formal writte ommunication like contracts among buyers and sellers, accounting reports, produc ion statistics, emails et cetera. Now it is easy to see that the group equilibriur nessage correspondence pu associates with each environment, 0, set of messages (0. hat are equilibrium messages for all the agents. Assume that the messages wer roposed actions, than j(@) consists of all proposals to which each agent would agre n 6. The outcome function h translates messages into outcomes. As we have see .mechanism 7 can be defined as an triple 7 = (M,,h). In the case 7 operates i . environment 6 than the result is outcome h(u(6)) in Z. In the given space 0, fo Jl environments, the mechanism 7 leads to the desired result by the agent in tha articular environment, than we say that 7 realizes F on ©. In short 7 realizes | f for all 6 in 6,h(w(0)) = F(@). Actually the equilibrium message ju represents th yehavior of the agents. The concept can be represented in a commuting diagram a hown in figure 2
ffirmative. The message exchange process consists of three elements first a messag pace M, second a group equilibrium message correspondence p, denoted uw: Or A nd third outcome function h, denoted ash: M+> Z. A message space M consist f the messages available for communication. Messages may include formal writte ommunication like contracts among buyers and sellers, accounting reports, produc ion statistics, emails et cetera. Now it is easy to see that the group equilibriur nessage correspondence pu associates with each environment, 0, set of messages (0. hat are equilibrium messages for all the agents. Assume that the messages wer roposed actions, than j(@) consists of all proposals to which each agent would agre n 6. The outcome function h translates messages into outcomes. As we have see .mechanism 7 can be defined as an triple 7 = (M,,h). In the case 7 operates i . environment 6 than the result is outcome h(u(6)) in Z. In the given space 0, fo Jl environments, the mechanism 7 leads to the desired result by the agent in tha articular environment, than we say that 7 realizes F on ©. In short 7 realizes | f for all 6 in 6,h(w(0)) = F(@). Actually the equilibrium message ju represents th yehavior of the agents. The concept can be represented in a commuting diagram a hown in figure 2
MODELING NORMS EMBEDDED IN SOCIETY Traditional the diagnostic problem is framed in situations where an observation of the system’s behavior is functioning abnormal or even fails to function at all. The issue is than to determine those components, objects et cetera of the system that will explain the difference between observed behavior and the desired correct behavior 62]. To solve the aforementioned diagnostic problem from first principles only the information of the system description is available together with the observation of the actual behavior. Reiter builds on the work of [41] and provides in a theoretical foundation for diagnosis from first principles. For representation purposes Reiter choose first-order logic for representing systems. As he observed and demonstrated many different logics lead to the same theory of diagnosis. Hence more abstractly Reiter’s theory can be formulated as a LDS and then translated into classical logic here first-order logic. In our situation there is one major difference and that is we cannot observe actual system behavior just because the system has to be designed yet. We do have a shared expectation about the expected behavior and we aim that the system after being build and implemented shows in practice the shared expected behavior. Indeed we have to consider that there is a possibility that the actua behavior after having the system built and implemented can actually differ from the expected outcome and we will need safeguards upfront to consider in designing the system. We coined this requirement incentive-compatible (direct or encoded revelation mechanisms. It needs no elaboration that the design problem and the diagnostic problems both share the same mechanisms and principles. This is easy to see in figure 4 [41].
MODELING NORMS EMBEDDED IN SOCIETY Traditional the diagnostic problem is framed in situations where an observation of the system’s behavior is functioning abnormal or even fails to function at all. The issue is than to determine those components, objects et cetera of the system that will explain the difference between observed behavior and the desired correct behavior 62]. To solve the aforementioned diagnostic problem from first principles only the information of the system description is available together with the observation of the actual behavior. Reiter builds on the work of [41] and provides in a theoretical foundation for diagnosis from first principles. For representation purposes Reiter choose first-order logic for representing systems. As he observed and demonstrated many different logics lead to the same theory of diagnosis. Hence more abstractly Reiter’s theory can be formulated as a LDS and then translated into classical logic here first-order logic. In our situation there is one major difference and that is we cannot observe actual system behavior just because the system has to be designed yet. We do have a shared expectation about the expected behavior and we aim that the system after being build and implemented shows in practice the shared expected behavior. Indeed we have to consider that there is a possibility that the actua behavior after having the system built and implemented can actually differ from the expected outcome and we will need safeguards upfront to consider in designing the system. We coined this requirement incentive-compatible (direct or encoded revelation mechanisms. It needs no elaboration that the design problem and the diagnostic problems both share the same mechanisms and principles. This is easy to see in figure 4 [41].
Related topics:
SociologyComputer ScienceMulti Agent System
580 California St., Suite 400
San Francisco, CA, 94104
© 2025 Academia. All rights reserved