Geber Ramalho

Recent advances on multi-agent patrolling

Patrolling is a complex multi-agent task, which usually requires agents to coordinate their decision-making in order to achieve optimal performance of the group as a whole. In previous work, many patrolling strategies were developed, based on different approaches: heuristic agents, negotiation mechanisms, reinforcement learning techniques, techniques based on graph-theory and others. In this paper, we complement these studies by comparing all the approaches developed so far for this domain, using patrolling problem instances that were not dealt with before (i.e. new map topologies). The final results constitute a valuable benchmark for this domain, as well as a survey of the strategies developed so far for this task.

Extracting and evolving mobile games product lines

For some organizations, the proactive approach to product lines may be inadequate due to prohibitively high investment and risks. As an alternative, the extractive and the reactive approaches are incremental, offering moderate costs and risks, and therefore sometimes may be more appropriate. However, combining these two approaches demands a more detailed process at the implementation level. This paper presents a method for extracting a product line and evolving it, relying on a strategy that uses refactorings expressed in terms of simpler programming laws. The approach is evaluated with a case study in the domain of games for mobile devices, where variations are handled with aspect-oriented constructs.

Multi-Agent Patrolling with Reinforcement Learning

Patrolling tasks can be encountered in a variety of real-world domains, ranging from computer network administration and surveillance to computer wargame simulations. It is a complex multi-agent task, which usually requires agents to coordinate their decision-making in order to achieve optimal performance of the group as a whole. In this paper, we show how the patrolling task can be modeled as a reinforcement learning (RL) problem, allowing continuous and automatic adaptation of the agentsý strategies to their environment. We demonstrate that an efficient cooperative behavior can be achieved by using RL methods, such as Q-Learning, to train individual agents. The proposed approach is totally distributed, which makes it computationally efficient. The empirical evaluation proves the effectiveness of our approach, as the results obtained are substantially better than the results available so far on this domain.

Bilateral Negotiation for Agent-Mediated Electronic Commerce

One of the main issues in electronic commerce is the inclusion of the negotiation facilities commonly available in human client-vendor interaction in real world commerce. E-commerce negotiation processes have usually been modeled as self-interested multi-agent systems. In these systems buyers and sellers are represented by agents that have opposite demands and decide what to do on the flight, based on the available information. However, currently ecommerce systems, such as Kasbah and Magma [21], provide a small number of bilateral negotiation facilities. Fortunately, some general negotiation models could be at first applied to e-commerce domain. This is typically the case of Faratins model, which can be seen as an extension of Kasbahs. In this paper, we propose an original bilateral agent negotiation model, which extends Faratins one. We introduce various facilities, such as alternative product suggestion, ultimatum generation, local contract agreements, etc. These facilities intend to grant users with a more flexible e-commerce environment. We present our model formalization, including the knowledge base that determines agent behavior. Some empirical validation is also presented to the case of computer purchase.

JEOPS - The Java Embedded Object Production System

Although the concept of intelligent agents facilitates the development of systems which require intelligent behavior, the lack of standard tools to support the creation of agents is still an obstacle for the broad use of this paradigm. Agents can be created via the use of either specialized languages or generalpurpose ones. We claim that the latter is more suitable for complex applications where components reuse is highly demanded. In this case, the integration of production rules and objects is a promising approach. This article presents JEOPS, an inference engine that aims to provide reasoning capabilities to Java. JEOPS gives a special attention to the problem of the rule-object integration uniformity, because of its impact on the system reusability and readability. JEOPS has been used in different projects, and the results are very encouraging.

Extracting and evolving code in product lines with aspect-oriented programming

For some organizations, the proactive approach to product lines may be inadequate due to prohibitively high investment and risks. As an alternative, the extractive and the reactive approaches are incremental, offering moderate costs and risks, and therefore sometimes may be more appropriate. However, combining these two approaches demands a more detailed process at the implementation level. This paper presents a method and a tool for extracting a product line and evolving it, relying on a strategy that uses refactorings expressed in terms of simpler programming laws. The approach is evaluated with a case study in the domain of games for mobile devices, where variations are handled with aspect-oriented constructs.

A Theoretical Analysis of Multi-Agent Patrolling Strategies

A group of agents can be used to perform patrolling tasks in a variety of domains ranging from computer network administration to computer wargame simulations. The quality of the strategies used for patrolling may be evaluated using different measures. Informally, a good strategy is one that minimizes the time lag between two passages to the same place and for all places. Recently, many different architectures of multi-agent systems have been proposed and evaluated on the patrolling problem [1]. In particular, different types of agents (reactive vs. cognitive), agent communication (allowed vs. forbidden), coordination scheme (central and explicit vs. emergent), agent perception (local vs. global), and decisionmaking mechanism (random selection vs. goal-oriented selection) were proposed. This paper proposes a theoretical analysis of the patrolling problem addressing the following issues : Are there efficient algorithms generating near-optimal strategies ? Are patrolling strategies based on partitioning the territory good?

Comparative analysis of porting strategies in J2ME games

Porting is a critical task in mobile device game development. The high diversity of devices requires correspondingly customized versions of a single game. Managing the variabilities among these customized versions while exploring the latent game commonality cannot be solely addressed with a single technique. This paper contributes to this task by identifying and analyzing porting challenges, and by evaluating and contrasting existing approaches within industrial-strength case studies of J2ME games. Further, we present lessons learned, proposing more effective guidelines for this process, aiming at improving the quality of the resulting applications and porting process productivity.

Negotiator agents for the patrolling task

Multi-agent systems can be used to perform patrolling tasks in various domains. In this work, we compare the results obtained by new negotiation based approaches with previous ones. By splitting the nodes of the world graph, the negotiator agents reduce the path they have to walk and the number of nodes to patrol, making it easier to maintain a low average idleness in world nodes. Auctions are the negotiation mechanisms used to split the nodes of the world, the agents bid on nodes based in their utility function. Empirical evaluation has shown the effectiveness of this distributed approach, as the results obtained are substantially better than those previously achieved by negotiator agents. The agent types presented in this work are more scalable and reactive since they can perform patrolling in worlds of all sizes and topology types. Besides, they are more stable as indicated by the low standard deviation obtained in node idleness.

Persona-AIML: an architecture for developing chatterbots with personality

This work presents the Persona-AIML architecture for the creation of chatterbots in AIML (Artificial Intelligence Markup Language) with personality. It is a flexible architecture that allows the use of different models of personality in the construction of chatterbots. Tests with the prototype revealed satisfactory and very encouraging results.