Agent Behavior Model Research Articles

Formal modeling and analysis of cognitive agent behavior

From an external perspective, cognitive agent behavior can be described by specifying (temporal) correlations of a certain complexity between stimuli (input states) and (re)actions (output states) of the agent. From an internal perspective the agent's dynamics can be characterized by direct (causal) temporal relations between internal and mental states of the agent. The latter type of specifications can be represented in a relatively simple, executable format, which enables different types of analysis of the agent's behavior. In particular, simulations of the agent's behavior under different (environmental) circumstances can be explored. Furthermore, by applying verification techniques, automated analysis of the consequences of the agent's behavior can be carried out. To enable such types of analysis when only given an external behavioral specification, this has to be transformed first into some type of executable format. An automated procedure for such a transformation is proposed in this paper. The application of the transformation procedure is demonstrated for a number of cases, showing examples of the types of analysis as mentioned for different forms of behavior.

A GIS based spatially explicit model of dispersal agent behavior

Spatial simulation models of seed dispersal have been constructed at the landscape level under the assumption of ubiquitous or uniform dispersibility. The anisotropic nature of vegetation distribution caused by different dispersal agents such as wind, gravity, water and animals were ignored. We propose a prototype of a GIS-based spatially explicit model of dispersal agent behavior (SEMODAR) to simulate the seed dispersal process by considering the unique behavioral characteristics of each seed dispersal agent. As a result, the influence of dispersal agent behavior on the species coexistence in competitive communities with and without habitat destruction could be explored. The model consists of four module components: dispersal rules, species competition, species colonization, and habitat destruction. An experimental simulation was conducted using three hypothetical species with differing competitive and migration abilities in both intact and disturbed conditions for 250 years. The findings of this study support the theoretical expectation that inferior competitors can coexist with superior competitors given that the inferior competitors have efficient colonization ability. The simulation also reveals the important role of agent behavior in the seed dispersal process and the biased impact of environment fragmentation on superior competitors that are not superior dispersers.

Arthur, George and the Mystery of the Missing Motive: Towards a Theory of Evidentiary Reasoning about Motives

Reasoning about motives is a prominent part of the investigative process in Julian Barnes' Arthur and George and in Conan Doyle's novels themselves. In Arthur & George, it is the lack of motive that originally convinces Arthur of George's innocence, a conclusion that a subsequent evaluation of George's character supports. In many of the Sherlock Holmes novels, it is the discovery of a hidden motive that puts Holmes on the right track. On closer inspection though, the evidentiary value and logical structure of arguments from motive turn out to be problematic. In this paper, we use concepts from argumentation theory and computational models of agent behavior from artificial intelligence research to analyze the structure of motive-based reasoning in fictional and factual crime investigations and trials. This analysis is used to develop a theory that (a) accounts for the rationality of motive- based reasoning, (b) helps to distinguish plausible and implausible arguments from motive, and (c) to distinguish arguments based on motive from arguments based on intention and character, as well as from other, legally more problematic forms of evidentiary reasoning.

Genetic algorithm based multi-agent system applied to test generation

Automatic test generating system in distributed computing context is one of the most important links in on-line evaluation system. Although the issue has been argued long since, there is not a perfect solution to it so far. This paper proposed an innovative approach to successfully addressing such issue by the seamless integration of genetic algorithm (GA) and multi-agent system. In the design phase, a test ontology was firstly defined for smoothing the communication among agents. For the implementation of GA, The fitness function and the structure of chromosome were identified on the basis of the analysis of constraint conditions associated with a test. To demonstrate the task execution flow and messages passing among agents, the activity diagram and sequence diagram were also shown on the AUML basis. In the phase of implementation, the JADE based agent behavior model was described in detail and the implementation platform was also demonstrated. The final simulation results validated the feasibility of the proposed approach.

Modeling mobile agent behavior

Mobile agents have been proposed for collecting and processing network management information in the Internet and other networks. This paper presents fundamental analytical stochastic models of mobile agent behavior for mobile agent design. The significance of these models is in their utility and flexibility. Models discussed include ones of agent dwell time, agent life span for various cloning scenarios, agent report statistics, and quality of service.reserved.

Agent behavior and agent models in unregulated markets

Mobile-agent systems show significant promise as the most effective way to harness the power of the Internet and the massive collection of information and opportunity that the Internet holds. However the efficient organization and control of these systems remains one of a number of unsolved problems with this approach to network computing. This paper examines a mobile-agent system with specific focus on environment sensing, preemptive load balancing and open agent markets. Agent behaviour is studied with actual agent systems using progressively sophisticated agent migration strategies.It is shown that actual modeling shows interesting and difficult to predict behaviour in the agent systems. It is shown that mobile agents with relatively simple migration strategies can cause loads in self-regulating agent markets to oscillate. It is further shown that using Autoregressive modeling to predict the market behaviour can allow individual agents to significantly outperform other agents. However the fidelity of the model is critical to the success of the agents. The criticality of good agent strategies and actual agent system modeling is thus highlighted.

How Common is Housing Discrimination? Improving on Traditional Measures

Fair housing audits are an important tool for measuring the incidence of racial and ethnic discrimination in housing markets. Traditional measures of incidence calculate the proportion of cases in which the white auditor is favored (gross adverse treatment) or the difference between the proportion of white-favored and minority-favored audits (net adverse treatment). The gross measure may overstate discrimination because treatment differences sometimes arise from differences in circumstances across auditor visits. On the other hand, the net measure understates discrimination because audits in which the minority auditor is favored for systematic reasons are incorrectly subtracted. This paper presents a model of agent behavior that more fully accounts for the audit design and employs the estimated parameters to calculate bounds on the incidence of discrimination. This approach leads to a lower bound for the incidence of discrimination that is often substantially higher than the simple net measure and to an upper bound that is close to and sometimes above the simple gross measure.