Caste-Centric Development of Agent-Oriented Information Systems

Based on the meta-model of information systems presented in Zhu (2006), this chapter presents a caste-centric agent-oriented methodology for evolutionary and collaborative development of information systems. It consists of a process model called growth model, and a set of agent-oriented languages and software tools that support various development activities in the process. At the requirements analysis phase, a modelling language and environment called CAMLE supports the analysis and design of information systems. The semi-formal models in CAMLE can be automatically transformed into formal specifications in SLABS, which is a formal specification language designed for formal engineering of multi-agent systems. At implementation, agent-oriented information systems are implemented directly in an agent-oriented programming language called SLABSp. The features of agent-oriented information systems in general and our methodology in particular are illustrated by an example throughout the chapter.

Solving Facility Location Problems with a Tol for Rapid Development of Multi-Objective Evolutionary Algorithms (MOEAs)

The low price of coffee in the international markets has forced the Federación Nacional de Cafeteros de Colombia (FNCC) to look for cost-cutting opportunities. An alternative that has been considered is the reduction of the operating infrastructure by closing some of the FNCC-owned depots. This new proposal of the coffee supplier network is supported by (uncapacitated and capacitated) facility location models that minimize operating costs while maximizing service level (coverage). These bi-objective optimization models are solved by means of NSGA II, a multi-objective evolutionary algorithm (MOEA). From a computational perspective, this chapter presents the multi-objective Java Genetic Algorithm (MO-JGA) framework, a new tool for the rapid development of MOEAs built on top of the Java Genetic Algorithm (JGA). We illustrate MO-JGA by implementing NSGA II-based solutions for the bi-objective location models.

Ant Colony Optimization and Multiple Knapsack Problem

The ant colony optimization algorithms and their applications on the multiple knapsack problem (MKP) are introduced. The MKP is a hard combinatorial optimization problem with wide application. Problems from different industrial fields can be interpreted as a knapsack problem including financial and other management. The MKP is represented by a graph, and solutions are represented by paths through the graph. Two pheromone models are compared: pheromone on nodes and pheromone on arcs of the graph. The MKP is a constraint problem which provides possibilities to use varied heuristic information. The purpose of the chapter is to compare a variety of heuristic and pheromone models and different variants of ACO algorithms on MKP.

How Genetic Algorithms Handle Pareto-Optimality in Design and Manufacturing

An informal analysis is provided for the basic concepts associated with multi-objective optimization and the notion of Pareto-optimality, particularly in the context of genetic algorithms. A number of evolutionary algorithms developed for this purpose are also briefly introduced, and finally, a number of paradigm examples are presented from the materials and manufacturing sectors, where multi-objective genetic algorithms have been successfully utilized in the recent past.

Genetic Algorithms for Organizational Design and Inspired by Organizational Theory

Modularity is widely used in system analysis and design such as complex engineering products and their organization, and modularity is also the key to solving optimization problems efficiently via problem decomposition. We first discover modularity in a system and then leverage this knowledge to improve the performance of the system. In this chapter, we tackle both problems with the alliance of organizational theory and evolutionary computation. First, we cluster the dependency structure matrix (DSM) of a system using a simple genetic algorithm (GA) and an information theoretic-based metric. Then we design a better GA through the decomposition of the optimization problem using the proposed DSM clustering method.

Population Symbiotic Evolution in a Model of Industrial Districts

This chapter considers a model of industrial districts where different populations interact symbiotically. The approach consists of the parallel implementation of the model with jESOF and plain C++. We consider a district decomposition where two populations, workers and firms, cooperate while behaving independently. We can find interesting effects both in terms of worker localization consequences and of the dynamic complexity of the model, with policy resistance aspects.By using a multiple implementation strategy, we compare the advantages of the two modeling techniques and highlight the benefits arising when the same model is implemented on radically different simulation environments; furthermore we discuss and examine the results of our simulations in terms of policy-making effects.

Modeling the Firm as an Artificial Neural Network

The purpose of this chapter is to make the case that first a standard artificial neural network can be used as a general model of the information processing activities of the firm; second, to present a synthesis of Barr and Saraceno (2002, 2004, 2005), who offer various models of the firm as an artificial neural network. An important motivation of this work is the desire to bridge the gap between economists, who are mainly interested in market outcomes, and management scholars, who focus on firm organization. The first model has the firm in a price-taking situation. We show that increasing environmental complexity is associated with larger firm size and lower profits. In the second and third models, neural networks compete in a Cournot game. We demonstrate that they can learn to converge to the Cournot-Nash equilibrium and that optimal network sizes increase with complexity. In addition, we investigate the conditions that are necessary for two networks to learn to collude over time.

Genetic Programming

The aim of genetic programming is to evolve programs or functions (symbolic regression) thanks to artificial evolution. This technique is now mature and can routinely yield results on par with (or even better than) human intelligence. This chapter sums up the basics of genetic programming and outlines the main subtleties one should be aware of in order to obtain good results.

Artificially in Social Sciences

This chapter provides an introduction to the modern approach of artificiality and simulation in social sciences. It presents the relationship between complexity and artificiality, before introducing the field of artificial societies which greatly benefited from the fast increase of computer power, gifting social sciences with formalization and experimentation tools previously owned by the “hard” sciences alone. It shows that as “a new way of doing social sciences,” artificial societies should undoubtedly contribute to a renewed approach in the study of sociality and should play a significant part in the elaboration of original theories of social phenomena.

Human Nature in the Adaptation of Trust

This chapter pleads for more inspiration from human nature in agent-based modeling. As an illustration of an effort in that direction, it summarizes and discusses an agent-based model of the build-up and adaptation of trust between multiple producers and suppliers. The central question is whether, and under what conditions, trust and loyalty are viable in markets. While the model incorporates some well-known behavioral phenomena from the trust literature, more extended modeling of human nature is called for. The chapter explores a line of further research on the basis of notions of mental framing and frame switching on the basis of relational signaling, derived from social psychology.