Dejan Mitrović

Radigost: Interoperable web-based multi-agent platform

Recent improvements of web development technologies, commonly referred to as HTML5, have resulted in an excellent framework for developing a fully-featured, purely web-based multi-agent platform. This paper presents an architecture of such a platform, named Radigost. Radigost agents and parts of the system itself are implemented in JavaScript and executed inside the clients web browser, while an additional set of Java-based components is deployed on an enterprise application server. Radigost is platform-independent, capable of running, without any prior installation or configuration steps, on a wide variety of software and hardware configurations, including personal computers, smartphones, tablets, and modern television sets. The system is standards-compliant and fully interoperable, in the sense that its agents can transparently interact with agents in existing, third-party multi-agent solutions. Finally, performance evaluation results show that the execution speed of Radigost is comparable to that of a non web-based implementation.

Extensible Java EE-Based Agent Framework – Past, Present, Future

EXtensible Java EE − based Agent Framework (XJAF) is a modular, FIPA-compliant multi-agent system developed by the authors of this chapter. The main motivation behind the development of XJAF was to demonstrate how existing, standardized Java EE technologies, tools, and libraries, such as JNDI, JMS, and EJB, can be used to implement a large subset of functionalities required from a multi-agent system. Immediate direct benefits of this approach are shorter development time of the system itself, delegation of agent load-balancing to the enterprise server, flatter learning curve for new developers of the system, etc. The first implementation of XJAF has been published several years ago and has since been used in several real-life applications. In the meantime, some disadvantages and weaknesses of the system were noticed, and the work is underway to provide a new implementation with an improved quality. The most recent focus of improvements has been on the addition of fault-tolerant techniques, and the increase of interoperability through a SOA-based design and web service interfaces.

Improving fault-tolerance of distributed multi-agent systems with mobile network-management agents

Large-scale agent-based software solutions need to be able to assure constant delivery of services to end-users, regardless of the underlying software or hardware failures. Fault-tolerance of multi-agent systems is, therefore, an important issue. We present an easy and flexible way of introducing fault-tolerance to existing agent frameworks. The approach is based on two new types of mobile agents that manage efficient construction and maintenance of fault-tolerant multi-agent system networks, and implement a robust agent tracking technique.

Introducing ALAS: A Novel Agent‐Oriented Programming Language

Agent‐oriented programming languages represent a family of programming languages that provide developers with high‐level abstractions and constructs necessary for implementing and using agent‐related concepts. In this paper a novel agent‐oriented programming language for rapid and efficient development of reactive agents, named ALAS, is presented. The simple, but powerful set of language constructs is designed to support the execution of agents in heterogenous environments, and to enable easy employment of advanced agent features, such as mobility and web service integration.

Agent-Based Distributed Computing for Dynamic Networks

Dynamic networks of personal computers are characterized by sudden, unexpected failures of computational nodes, as well as the addition of new computers at any given moment. This paper presents a new agent-oriented system for distributed computing, named ADiS, specifically designed for operating in these kinds of environments. The system relies on the agent technology in order to adapt to dynamic changes in a computational net-work. Originally envisioned as a distributed system for distance matrix calculations, ADiS has recently been redesigned as a general-purpose, extensible architecture for arbitrary computing in a distributed environment. Experimental results demonstrate that the usage of ADiS has a significant positive impact onto the execution time of lengthy processes. DOI: http://dx.doi.org/10.5755/j01.itc.43.1.4588

Different Roles of Agents in Personalized Programming Learning Environment

Researchers had recently begun to investigate various techniques to help learners to improve learning effects using e-learning systems. In this paper, we propose an e-learning architecture with a recommendation module consisting of several different kinds of pedagogical agents which actively participate in learning processes, provoking learners and motivating them to learn more effectively. Different kinds of agents can be introduced in order to support scaffolding activities in learning programming languages and problem solving.

Delivering the multiagent technology to end-users through the web

HTML5 represents an important milestone for the modern web. It offers a standardized way of developing feature-rich Internet applications in a platform-independent manner. This paper presents a software platform for deploying purely web-based agents. It provides the necessary infrastructure for utilizing many techniques of the multiagent technology, such as agent interaction and mobility, in modern web applications.

Distributed distance matrix generator based on agents

The process of calculating similarities between different time series of a dataset -- i.e. a distance matrix -- is often very time-consuming, but can be sped-up significantly using distributed programming. This paper presents a distributed system for calculating the distance matrix that utilizes the agent technology. Software agents are employed to deal with dynamic properties of the computational network, such as sudden unavailability of a number of computers, as well as for load-balancing and task-sharing.

Agent-based approaches to managing fault-tolerant networks of distributed multi-agent systems

Large-scale agent-based software solutions need to be able to assure constant delivery of services to end-users, regardless of the underlying software or hardware failures. Fault-tolerance of multi-agent systems is, therefore, an important issue. We present two algorithms for an easy and flexible introduction of fault-tolerance to existing agent frameworks. The first algorithm is based on a new type of mobile agent, named Connection Agent, for efficient construction and maintenance of fault-tolerant multi-agent system networks. The algorithm has been experimentally verified, and then significantly optimized by relying on the mobility feature of Connection Agents. Secondly, a robust agent tracking technique based on a special type of agent, named RemnantAgent, is proposed.

An Approach of Temporal Difference Learning Using Agent-Oriented Programming

Reinforcement Learning -- RL is an important agent problem that was not approached using the tools provided by Agent Oriented Programming -- AOP. Agent Speak (L) and its Jason implementation based on Java platform are representing state-of-the-art approaches of AOP based on the Belief-Desire-Intention -- BDI model. Temporal Difference Learning -- TDL is a passive RL method that can be used by an agent to learn its utility function while it is acting according to a given policy in an uncertain and dynamic environment. In this paper we present an approach for modeling and implementation of TDL using the Jason AOP language. So, our paper is presenting a contribution towards narrowing the gap between RL and AOP, by endowing BDI agents with TDL skills.