Agent and Multi-Agent Systems. Technologies and Applications

Many important developments in artificial intelligence have been stimulated by organized competitions that tackle interesting, difficult challenge problems, such as chess, robot soccer, poker, robot navigation, stock trading, and others. Economics and artificial intelligence share a strong focus on rational behavior. Yet the real-time demands of many domains do not lend themselves to traditional assumptions of rationality. This is the case in many trading environments, where self-interested entities need to operate subject to limited time and information. With the web mediating an ever broader range of transactions and opening the door for participants to concurrently trade across multiple markets, there is a growing need for technologies that empower participants to rapidly evaluate very large numbers of alternatives in the face of constantly changing market conditions. AI and machine-learning techniques, including neural networks and genetic algorithms, are already routinely used in support of automated trading scenarios. Yet, the deployment of these technologies remains limited, and their proprietary nature precludes the type of open benchmarking that is critical for further scientific progress.

The Trading Agent Competition was conceived to provide a platform for study of agent behavior in competitive economic environments. Research teams from around the world develop agents for these environments. During annual competitions, they are tested against each other in simulated market environments. Results can be mined for information on agent behaviors, and their effects on agent performance, market conditions, and the performance and behavior of competing agents. After each competition, competing agents are made available for offline research. We will discuss results from various competitions (Travel, Supply-Chain Management, Market Design, Sponsored Search, and Power Markets).

Cooperation is the fundamental underpinning of multi-agent systems, allowing agents to interact to achieve their goals. However, agents must manage the risk associated with interacting with others who have different objectives, or who may fail to fulfill their commitments. There are many ways in which such a desirable social order may be encouraged or even mandated. For example, trust offers a mechanism for modeling and reasoning about reliability, honesty, etc., while organisations and norms provide a framework within which to apply them, and motivations provide a means for representing and reasoning about overall objectives. In this talk, I will consider the role of all these aspects, with a particular focus on norms in regulating behaviour.

Networked swarms may be considered as heterogeneous, distributed agent-based systems (or distributed wireless sensor networks), where the concept/definition of an ’agent’ may be interpreted in the most general sense. This is true for networked swarms of unmanned systems, as well as manned-unmanned systems, where the term ’agent’ may be thought of as (an individual) ’asset’. When considering swarms in (fixed or loose) 2-D/3-D formations, each swarm may be also thought of as a ’group of agents’ with initially specific, but dynamically and in real-time modifiable goals pertaining to an assigned mission or missions. As such, when studying such swarms, there are major challenges related to navigation and control, communications (including secured inter- and intra- swarm communication), coordination and collaboration, robustness and fault tolerance, security, to name just a few.

In this seminar, the ’connection’ between the swarm formation consideration and the ’distributed (agent-based) sensor network’ consideration will be established. Assets will be considered as sensor nodes, and the overall swarm as a sensor network. A well integrated methodology will be presented, in which on top of navigation/control aspects, communication and networking aspects as well as software security will be discussed. The swarm formation will be dictated / modified based on communication/data transmission requirements, in addition to control requirements. Further, it will be shown how networked swarms / distributed agent-based systems may be organized in architectures comprised of swarms, each containing a ’swarm head’ as leader with all ’swarm heads’ being connected via a backbone network.

Case studies of aerial, ground and aerial-ground unmanned vehicle swarms will be demonstrated experimentally.

In this paper, the decentralized mobile target allocation problem is researched. We assume the existence of two groups of mobile agents: attackers and targets. Every attacker agent gets allocated to its best target based on the communication and coordination with the rest of the group positioned within a limited communication range (radius) and moves towards it. This is performed through the dynamic iterative auction algorithm with mobility without any insight in the decision-making processes of the other agents. Targets are mobile and combine two strategies to escape from the attacking group: moving linearly and randomly away from the attacker. We explore the dynamics of the allocation solution in respect to the mentioned escape strategies, maximum step size in each run (velocity), and the agents’ varying communication range when the latter is not sufficient to maintain a connected communication graph.

The paper describes design and implementation of Bayesian proprioceptor for forest fire observer network. The proprioceptor, sometimes also referred to as network observer has task of syintactical and semantical sensor and data validation in advanced sensor network Multi agent Bayesian network is used for cooperative data analysis and data understanding having false alarm reduction as final goal. A multi agent system for data sampling and data analysis is described. The proprioceptor is deployed as a part of intelligent forest fire monitoring system (iForestFire).

The main aim of this paper is to point out that the agent cooperation based on supervision can be utilized also in crisis situations. The approach is applied to an evacuation of an endangered area (EA). The supervisors for agents modelling crisis situations are synthesized and presented here. This makes possible not only the agent cooperation in itself but also to check the evacuation process and guarantee full evacuation of people from EA. Place/transitions Petri Nets (P/T PN) are utilized in order to model the elementary autonomous agents being modules of EA. The supervision of agents is based on the DES (discrete-event systems) control theory. It is realized by means of known DES supervision methods. The proposed approach is illustrated in details on the supervision-based cooperation of agents in a case study of evacuation. The corresponding evacuation workflow is proposed by means of PN too. Moreover, the supervision is applied also in case of the workflow.

By wrapping services as active software agents, constantly bidding to consumers’ requests, self-organizing web service compositions can become real. But they still lack a negotiation capability. In this paper, we add to our Multiagent component composition system (MACOCO) the negotiation behavior of non-functional requirements, by implementing a modified version of the Zeuthen strategy, which allows both parties to get a fair agreement without knowing the utility function of the counterpart. We study the tradeoff of extra overhead versus a fair agreement, showing the feasibility and performance of this approach. A web prototype tool is available.

The modern teaching methods should allow effective learning in a short time. Development and detailed analysis of an e-learning system as well as a research on influence of the applied didactic methods on effectiveness of the learning process is required. Therefore, this paper is devoted to presentation of the results of an experiment conducted using prepared prototype of an e-learning system with a personalized courseware. Authors run researches on an influence of a personalization on the learning process effectiveness. Significantly higher results were obtained by learners whose user profiles were taken into consideration during the determination of the learning scenario. This conclusion is true when the learning material is presented for the first time, when users are underaged, male or have had elementary education.

In complex application systems, there are typically not only autonomous components which can be represented by agents, but humans may also play a role. The interaction between agents and humans can be learned to enhance the stability of a system. How can agents adopt strategies of humans to solve conflict situations? In this paper, we present a learning algorithm for agents based on interactions with humans in conflict situations. The learning algorithm consists of four phases: 1) agents detect a conflict situation, 2) a conversation takes place between a human and agents, 3) agents involved in a conflict situation evaluate the strategy applied by the human, and 4) agents which have interacted with humans apply the best rated strategy in a similar conflict situation. We have evaluated this learning algorithm using a Jade/Repast simulation framework. An evaluation study shows two benefits of the learning algorithm. First, through interaction with humans, agents can handle conflict situations, and thus, the system becomes more stable. Second, agents adopt the problem solving strategy which has been applied most frequently by humans.

This paper presents implementation of an algorithm that handles knowledge management and learning processes which are implemented in the existing Multi-Agent System (MAS). Both presented knowledge management and learning mechanisms are based on the implemented ontology mechanisms. Proposed MAS solution is based on the Java Agent Development (JADE) framework which is a Java based tool that supports creation of FIPA (Foundation for Intelligent Physical Agents) compliant agents. It integrates with various different OPC (OLE for Process Control) and database servers simultaneously. OPC is an open automation communication interface that is used to retrieve and analysis of various different real time data from the various different remote process controllers. To integrate OPC automation interface, presented MAS utilize Java Native Interface (JNI) which is a bridge between a Java and native programming languages. Presented MAS establishes database cooperation by means of the NHibernate entity framework which is a .Net platform specific, object relational mapper (ORM). Integration with .Net platform is performed by means of the highly refactored JADE Sharp module add-on which enables creation .Net agents compliant with the JADE framework.

This publication proposes using biological modelling methods based on DNA cryptography to semantically analyse data. The solutions proposed are illustrated with the example of E-UBIAS systems which analyse image data in combination with the identity analysis (personalisation of individuals). The use of DNA cryptography to analyse data makes it possible to unanimously assign analysed data to an individual at the personal identification stage. At the same time, supplementing the system with semantic analysis processes conducted based on semantic interpretation allows the possible lesions that the person suffers from to be identified.

The hierarchical data format becomes increasingly common. One of the problems arising with it is efficient tree integration, i.e. integration of XML databases. One possible approach is path-based integration, used among others in applications using XQuery and XPath database queries. This article offers a generalization for integration tasks based on this approach, as well as a short analysis of the result in terms of other existing integration criteria. The paper presents estimations for such criteria as Completeness, Minimality and Precision.

Appropriate data mining exploration methods can reveal valuable but hidden information in today’s large quantities of transactional data. While association rules generation is commonly used for transactional data analysis, clustering is rather rarely used for analysis of this type of data. In this paper we provide adaptations of parameters related to association rules generation so they can be used to represent distance. Furthermore, we integrate goal-oriented quantitative attributes in distance measure formulation to increase the quality of gained results and streamline the decision making process. As a proof of concept, newly developed measures are tested and results are discussed both on a referent dataset as well as a large real-life retail dataset.

In order to achieve flow (i.e. complete focus on playing followed by a high level of enjoyment) and increase player retention (i.e. keep a user playing a game longer and more often) it is important that difficulty of the game that a user is playing matches her/his skills. Due to a large amount of different games which are available to users, it is not easy for them to find games which best suit their skills and abilities. In this paper we propose a recommendation algorithm based on the information gathered from users’ interaction with a game. We use that information to model users’ success and progress in the game as well as motivation for playing. Besides, the proposed algorithm also takes into account user preferences, mobile phone characteristics and game related information which is gathered from users once the game is available on the market. Before enough information is gathered from users, the algorithm uses the information gathered during the game development phase and acquired from game developers and testers. In the implemented multi-agent system, after a user finishes playing a game, she/he receives a notification with a list of games which best suit her/his skills and preferences.

A feature of Croatian and other Slavic languages is a rich inflection system, which does not exist in English and other languages that traditionally dominate the scientific focus of computational linguistics. In this paper we present the results of the experiments conducted on the corpus of the Croatian online spellchecker Hascheck, which point to using non-nominative cases for discovering collocations between two nouns, specifically the first name and the family name of a person. We analyzed the frequencies and conditional probabilities of the morphemes corresponding to Croatian cases and quantified the level of attraction between two words using the normalized pointwise mutual information measure. Two components of a personal name are more likely to co-occur in any of the non-nominative cases than in nominative. Furthermore, given a component of a personal name, the conditional probability that it is accompanied with the other component of the name are higher for the genitive/accusative and instrumental case than for nominative.

LCC is a Lightweight Coordination Calculus which can be used to provide an executable, declarative specification of an agent interaction model. In this paper, we describe an LCC-based system for specifying the migration behaviour of virtual machines in a datacentre. We present some example models, showing how they can be used to implement different policies for the machine allocation and migration. We then describe a practical implementation of the system which can directly execute the LCC specifications.

The upsurge of Micro-blogging platform attracts enterprises to use it as a public relationship tool. It also act as a new form of news source, journalists can hunt for next upcoming breaking news. It is worth to identify communities from it and reveal social relationships among community members in a timely manner. However, traditional SNA approaches are insufficient to achieve the requirement in a reasonable time. In this paper, we proposed a hybrid framework to tackle the problem. It is designed to identify the community with real social relationships automatically, that withstand dynamically changing content, have the ability to process fast and live-streaming data and provide a self-feedback mechanism to refine the result without human interference. The benefit of this framework is that average users should be able to employ it and to really understand communities in micro-blogging platforms without any or limited domain knowledge.

In data intensive workflows, which often involve files, transfer between tasks is typically accomplished as fast as the network links allow, and once transferred, the files are buffered/stored at their destination. Where a task requires multiple files to execute (from different previous tasks), it must remain idle until all files are available. Hence, network bandwidth and buffer/storage within a workflow are often not used effectively. In this paper, we are quantitatively measuring the impact that applying an intelligent data movement policy can have on buffer/storage in comparison with existing approaches. Our main objective is to propose a metric that considers a workflow structure expressed as a Directed Acyclic Graph (DAG), and performance information collected from historical past executions of the considered workflow. This metric is intended for use at the design-stage, to compare various DAG structures and evaluate their

potential

for optimisation (of network bandwidth and buffer use).

We created a set of domain ontologies that are based on user requirements for the Fish4Knowledge (F4K) project – goal, video description and capability. The roles of the ontologies are to 1) support the development of appropriate functions of the project’s workflow system, and 2) serve as a communication media to interface with other F4K components. The ontologies were designed with collaboration with image processing experts, marine biologists and user interface experts to capture the domain knowledge succinctly. They were utilised in the first version of our workflow composition and execution system for video classification, fish detection and counting tasks. They will continue to evolve with F4K’s needs and are envisaged to interface with other components.

Cloud computing promises efficient use of hardware resources through virtualization and elastic computing facilities. Various cloud computing solutions have emerged on the market from open-source communities and commercial vendors. In this paper we discuss criteria for feature comparison of private cloud platforms and compare several open-source and commercial products. We test performance of hypervisors used in these clouds with a set of benchmark suites containing tests for various aspects of the system. We discuss the results in the context of what is commonly described as a scientific workload. The described feature and performance differences can help make wiser platform choices.

This paper presents a MAS dedicated to abnormal behaviors detection and alerts triggering in the maritime surveillance area. This MAS uses anomalies issued from an experienced Rule Engine implementing maritime regulation. It evaluates ships behavior cumulating the importance of related anomalies and triggers relevant alerts towards human operators involved in maritime surveillance. These human operators evaluate triggered alerts and confirm or invalidate them. Invalidated alerts are sent back to the MAS for a learning step since it self-adapts anomalies values to be consistent with human operators feedbacks. This MAS is implemented in the context of the project I2C, an EU funded project dedicated to abnormal ships behavior detection and early identification of threats such as oil slick, illegal fishing, or lucrative criminal activities (e.g. goods, drugs, or weapons smuggling).

This paper describes a system that can efficiently control the motion of automated guided vehicles (AGVs) in decentralized autonomous flexible manufacturing systems (DA-FMS). In particular, a mind model called the minimum unit of the mind (MUM) is proposed. Because of the change of mind characteristic of MUM, the AGVs can efficiently avoid route interferences even when many AGVs are operating in DA-FMS. The components of MUM are two stimulation cells, a box, and an arrow. Using application simulations, it is ascertained that MUM is effective for avoiding AGV route interferences.

In this work we discuss the advances in the development of an existing agent platform [20] taking in mind the domain of intelligent surveillance. We have paid special attention to key aspects such as scalability, robustness and flexibility. Furthermore, we have focused our effort to i) implement a scalable replication service for the main components of the platform, ii) provide the developers with a flexible agent-based development model and iii) integrate concurrent negotiations. A MAS deployment is also described to monitor an indoor environment.

Application of the Agent and Multiagent Systems (AMAS) in the industrial continuous processes can be a quite interesting and effective solution, especially for monitoring and controlling purposes of bioreactor systems. In the classical approach, a process operator controls the process, but sometimes must take some essential decisions concerning the choice of control strategy. In the case of biological processes, due to their highly nonlinear nature, this can be quite difficult task. For instance, the oscillatory behavior of the bioreactor may lead to higher or lower average biomass concentrations. Hence, there is a need to support the operator by measuring and controlling some additional parameters and this cannot be achieved using only measuring devices and classical control algorithms alone. Based on the agent technology, it has been shown that it is possible to support the operator and to achieve process goals.

Persons with complex communication needs are individuals who are unable to use speech as their primary method of communication. Reasons for complex communication needs differ widely, such as physical and/or intellectual impairment, Down syndrome, Autistic Spectrum Disease etc. Alternative and augmentative communication (AAC) is based on the use of graphic and textual symbols to represent certain objects, actions, or concepts. AAC has been proven as highly beneficial for improving speech, literacy, learning, employment, and quality of life for persons with complex communication needs. ICT-based AAC services refer to services that enable symbol-based human-to-human and human-to-machine communication in a computer and networking environment.

All children with complex communication needs have the right to engage in interpersonal communication and participate in society. The field of augmentative and alternative communication (AAC) provides education, systems and practice that enable such participation. AAC requires a multidisciplinary approach. A speech and language pathologist (SLP) is an important member of the team responsible for assessment, implementation and evaluation within AAC. This paper presents the current knowledge and practice of SLPs in Croatia in the area of AAC. The data shows that SLPs in Croatia rarely use aided and high-technology tools with clients. Three cases of children with low levels of expressive communication are elaborated from the aspect of their needs and intervention. It is concluded that the area of AAC has a great need for attention in the following: 1) further promotion among the general public, 2) multidisciplinary collaboration and 3) additional education for professionals.

Intensive development of information and communication technologies can contribute greatly to enhance alternative and augmentative communication for individuals who experience difficulty communicating in an understandable manner. Availability of internet infrastructure and affordability of mobile computer devices provide a sound starting point for planning and building AAC services. In order to make the services more familiar and understandable to users, it is justifiable to build them on top of common service platforms in an adaptive and configurable manner. In this paper we present a component-based AAC service development model that enables building such services by using mandatory AAC components on top of the component framework, and selecting visual or non-visual components from the available component pool. The AAC component selection criteria are based on matching the data and metadata of usability requirements specified with the metadata of available components from the component pool. Registering new components into the reusable component pool extends the pool and helps building the component-based AAC services development platform.

Augmentative and Alternative Communication (AAC) includes various communication methods used by people with speech and language impairments, and is often aided by electronic devices such as smartphones or tablets. Since skills and limitations vary for each user, it is possible to create an interface on these electronic devices that would adjust to user’s needs in ways that would significantly improve usability and ease of use. Paper describes e-Accessible service system containing Calibrator and Communicator services. Calibrator measures user’s preferences and limitations regarding user interface through calibration process and calculates UI parameters adapted to the particular user. The parameters are forwarded to the Communicator service, whose main purpose is to supplement speech or writing for people with complex communication needs.

Rapid evolution of mobile devices enables the today’s user to access the content and technology previously reserved for multiple devices in one compact, portable package. The introduction and popularization of smartphones, tablets and e-readers changed the way that the users communicate, consume and create content. This paper examines the potential of using such devices for symbol-based Augmentative and Alternative Communication. We have proposed a model of an adaptive symbol-based AAC application for mainstream mobile devices. In respect of the requirements of the model we have examined technical possibilities and limitations of specialized and mainstream tablet devices available. Via a preliminary research of potential symbol-based AAC users in Croatia we have investigated the current state, needs and requirements regarding the use of digital AAC applications.

In the modern society, inclusion of persons with disabilities in the mainstream education is increasingly enhanced. The presented example focuses on inclusion of individuals for whom their learning environment is created and adapted according to their special needs and capabilities. In establishing the learning environment, the role of multimedia and ICT is important. Persons with disabilities are often disadvantaged in ICT and in participating in the classroom collaborative activities. The paper presents a novel interactive application utilizing tangible user interface (TUI) as an alternative input device for computer-supported collaborative learning and game based-learning. In the final part of our paper, we summarize the main findings and results of evaluation of our interactive game for the ICT-assisted learning of people with special needs often having them tested in a real educational environment. The findings indicate high usability of TUI for people with physical impairment and severe to mild learning difficulties for whom it enables inclusive ICT and e-accessibility.

This paper presents the first results achieved within the Indiana MAS project funded by Italian Ministry for Education, University and Research, MIUR. We discuss how the AgentSketch holon belonging to the Indiana MAS has been extended to cope with images, besides hand drawn sketches, and has been tested in the domain of Mount Bego’s prehistoric rock art (southern French Alps). The way Indiana MAS holons cooperate in order to provide correct interpretations of ambiguous shapes is discussed by means of an example based on hypotheses recently advanced by archaeologists.

This paper proposes an approach that makes it possible to have a robust multi agents system. Generally, a multi agents system has a mission called the global goal that it has to achieve. This global goal can be decomposed into different sub goals. We define two types of goals: the simple sub goal realized by only one agent and the complex sub goal achieved by more than one agent. The agents that realize simple goals formed one class, and they are replicated to assure fault tolerance. Each group of agents in cooperation formed one group called the complex goal class; it uses the exception handling technique to treat the errors. We use extern agents to manage the classes formed and to control the whole system.

This paper documents the case to address the distributed data storage and widely used unique identifiers which are in use from the beginning of databases and backend systems. The importance of uniqueness for object system identifiers is crucial for efficient data retrieval, storage and comparison. We are representing the new method for objects tagging using synergy mechanism between well known Geohash algorithm and Universally Unique Identifiers we call GHUUID (Geohash+UUID) gaining the agent system spatio-temporal identification of all data available. Multi-Agent systems in their nature strongly rely upon distributed databases across vast amount of data. Some of the interoperability issues are solved with usage of semantic web principles which can be cumbersome to implement and maintain. Most systems need the basic space-time correlation which can be the corner stone for the efficient analysis and data aligning regardless of objects stored and their nature.

The techniques of multi-agent system bring intelligence and flexibility to embedded agent/multi-agent embedded system connected to internet presents a great advantage. Advantage of such multi-agent systems improves the use of expanded infrastructure. Installing simplified agents in embedded systems was shown necessary due to increasing use of embedded devices. Multi-agent power management system is based on battery control and anticipation of replacement. Wireless transmission of ECG (electrocardiogram) signal via ZigBee (XBee modules) brings some problem into focus. This paper presents concept realized and tested on real equipment. Using smart mobile phone (today a widely used device) interaction/actoric between end user embedded agent and embedded master agent can give feedback about end users health in real-time. Similar, off-line monitoring device exists, but not connected to the network (Holter). Monitoring of patients in real-time can be enabled by such device that exhibits wireless communication and allows transmission of real-time source signal. The agent on which this paper refers to is program/firmware function in program code in small embedded system for monitoring ECG signal. Simplified agent activates the watchdog for battery alert. Main focus is to control the power consumption of WSN (Wireless Sensor Node). Power management has the intelligence middleware and allows timely to respond and inform the end users. Artificial intelligence is integrated in master agent that is element of embedded system cloud and a primal high level layer. Secondary layer is integrated in dedicated servers which respond to device clouds. LabVIEW application for signal processing provides robust and efficient environment for resolving ECG signal processing problem. These tool/application can be also used in other biomedical signal processing applications such as Magnetic Resonance Imaging (MRI) and Electroencephalography (EEG).

Integrated process planning and scheduling (IPPS) is an NP-hard problem, the major research on the multi-agent system (MAS) based IPPS systems has focused on the establishment of negotiation protocols to accomplish the integration of process planning and scheduling. However, not much consideration has been paid to the dynamic factors of current manufacturing systems. In this paper, an MAS architecture is proposed to solve the dynamic IPPS problem with embedded heuristic algorithms. The proposed MAS system can be combined with a variety of heuristic methods to support dynamic process planning, scheduling and re-scheduling. As a result, the proposed MAS system for dynamic IPPS using heuristics possesses high flexibility, extensibility, and accessibility for manufacturing applications.

Distributed constraint optimization problems have attracted attention as a means of resolving distribution problems in multiagent environments. The author has already proposed a multiplex method targeting the improved efficiency of a distributed nondeterministic approximate algorithm for distributed constraint optimization problems. Since much of the computation time is used to transmit messages, improving efficiency using a multiplex computation of distributed approximate algorithms might be feasible, presuming that the computation time of each node or a small change in message length has no direct impact. Although it is usually impossible to guarantee that the approximation algorithm can obtain the optimal solution, the author managed to do so, using a theoretically determined multiplex method. In addition, the author shows the feasibility of an optimal solution attainment rate of 0.99 by an experiment using a Distributed Stochastic Search Algorithm.

We propose a MDA based approach for designing secure mobile agent systems. First, we define a meta-model which extends the UML deployment diagram by concepts related to the security and mobility of multi-agent systems. We propose also a UML profile as an implementation of this meta-model. Second, we project the application model into AGLETS-specific model, which describes the main functionalities of the application deployed on AGLETS as a mobile agent platform.

The paper deals with modeling and simulation of business processes. A multiagent system was implemented as a tool to manage the simulation. Multiagent systems often operate with random (respectively pseudorandom) generated parameters in order to represent unpredictable phenomena. The aiml of the paper is to show the influence of different random number generation functions to the real multiagent system outputs. It is obvious, that outputs of the multiagent system simulation differs from turn to turn, but the motivation was to find, if the differences are significant. An accurate number of agents with the same parameters were used for each case, with different kinds of randomness while generating agent’s internal state attributes. The results obtained show that using inappropriate random number generation function leads to significant output data distortion, so the generation function selection must be done very carefully.

We provide a simple computational structure for representing recursive

mental states

that could constitute the semantic level of a formal language to deal with the dynamics of mental states. For the sake of simplicity, we base our conception of mental state on only two primitive attitudes:

beliefs

(for the cognitive sphere) and

desires

(for the volitional sphere).

The paper is devoted to a novel formalization of beliefs in multiagent systems. Our aim is to bridge the gap between idealized logical approaches to modeling beliefs and their actual implementations. Therefore the stages of belief acquisition, intermediate reasoning and final belief formation are isolated and analyzed. We give a novel semantics reflecting those stages and suitable for building complex belief structures in the context of incomplete and/or inconsistent information. Namely, an agent starts with constituents, i.e., sets of initial beliefs acquired by perception, expert supplied knowledge, communication with other agents and perhaps other ways. Next, the constituents are transformed into consequents according to agents’ epistemic profiles.

Additionally, a uniform treatment of single agent and group beliefs is achieved. Importantly, we indicate an implementation framework ensuring tractability of reasoning about beliefs.

Production systems must be reconfigurable and endowed with methods and tools allowing an automatic recovery of expected and unexpected errors. In this domain, error recovery in production systems is always an open problem. Multi-agent systems seem to be adapted for representing this kind of problem in an attempt to arrive at a reconfigurable, adaptive, and “intelligent” manufacturing systems. The objective of this work consists in proposing a specification, modeling and structural analysis approach of error recovery production systems using holonic multi-agent systems and Object Petri Nets (OPN).

A multi-agent system consists of a group of interacting autonomous agents. The key problem in such a system is coordination and cooperation, i.e. how to ensure that individual decisions of the agents result in jointly optimal decisions for the overall system. This problem becomes even more serious when the number of the agents is large. Holonic model is an effective method to manage large scale problems. In holonic approaches, the formation of the initial holons is very critical and has a great influence on their performance and effectiveness. In this paper, we use a graph based modelling approach to group a population of agents with a greedy method, driven by a very simple and effective quality measure. The proposed method is evaluated by applying it to an urban traffic problem as a case study and it is shown the proposed method produces better results.

Many deaf people are using lip reading as a main communication fiorm. A viseme is a representational unit used to classify speech sounds in the visual domain and describes the particular facial and oral positions and movements that occur alongside the voicing of phonemes. A design tool for creating correct speech visemes is designed. It’s composed of 5 modules; one module for creating phonemes, one module for creating 3D speech visemes, one module for facial expression and modul for synchronization between phonemes and visemes and lastly one module to generate speech triphones. We are testing the correctness of generated visemes on Slovak speech domains. The paper descriebes our developed tool.

This paper presents a cost based probability updating technique based on Bayesian decision theory for modifying decision confidences. The technique allows for the incorporation of prior knowledge via the use of cost matrices supplied by another source of intelligence. Data signals acquired by a metal detector array corresponding to UXO based targets were used to evaluate the technique with the assistance of a previously developed automated decision system. The classification classes utilised were based on target depth level and metal type. The results showed that the probability updating technique was able to produce an increase in the classification performance and also reduce the classification errors below approximately 5 to 10%.

A methodology is presented for the assessment of human operator performance in a detection and identification task, using two sets of infrared images of natural outdoor scenes with everyday objects used as targets. It includes measures of effectiveness such as operator detection rate, identification rate, false alarm rate, response time, confidence levels and image quality ratings. This robust methodology could be used in the evaluation of any image improvement technique or to evaluate different imaging techniques or technologies.

In this paper we describe modeling for verification of business process with Spin model checker. Our primary goal is the development of Promela language description for e–invoice web service. Modeling for verification follows Church’s synthesis problem: for input scenario “data–flow” model, output is Promela model. Sequence of model transformations translates the scenario into the Promela language code. At the end whole process is illustrated with e–invoice web service example.

The technologic processes are executed in the recent technological systems by process directing devices. In these systems the instruments fix the measured values and the corresponding dates (timestamps) on the one hand, on the other hand it is also recorded who and what performed what kind of intervention during operation. The goal of our research is to work out special agents for resulting data from monitoring of functioning of discrete event systems with special regard to diagnosis of faulty mode of operation. In our research we examined vehicle industry as to what useful information can be filtered out for the sake of functional and cost efficiency from the logs resulting from the processes. One of the areas is one of the testing processes in vehicle industry which follows planning, and the other one is the examination of specific troubleshooting process in case of buses.

Conventional pos-neg model of sentiment analysis primarily for review documents is inappropriate for news articles because of the sentiment diversity of the latter. We design three-dimension sentiments that are more suitable for the analysis of news articles. For a contentious topic, different news websites may have different sentiment tendencies and the tendencies may vary over time. To catch this feature, we construct a sentiment dictionary and develop a system that can extract news articles’ sentiments, present sentiment variation over time inside a news website, and compare sentiment correlation between news websites.

In this paper the evolution of

EXtensible Java EE-based Agent Framework

(

XJAF

), a

FIPA

-compliant multi-agent system will be presented. The main motivation behind the development of

XJAF

is to demonstrate how existing, standardized

Java EE

technologies, libraries and technical solutions (e.g.

JNDI

,

JMS

, and

EJB

), can be used to implement a large subset of functionalities required by a multi-agent system. Immediate benefits of this approach are shorter development time, harnessing of advanced programming features such as agent pooling and runtime load-balancing, flatter learning curve for new developers of the system, and so on. Several weaknesses that became apparent with the original implementation of

XJAF

caused recent improvements of the system. These include the addition of fault-tolerant techniques, and the increase of interoperability through a

SOA

-based design.

With intense increase in number of competing service providers in the information and communication sector, companies must implement mechanisms for forecasting consumer interest in new services. Common growth models provide the mechanisms for modelling and predicting acceptance of a certain service. However, they have two shortcomings: i) limited precision; and ii) a short, but yet existing, time delay. By using semantic reasoning for detecting similarities between services already on a market and ones that are just to be introduced, it is possible both to increase forecasting precision and eliminate the time delay caused by the need to collect a certain amount of data about the new service before a prediction can be made. The proposed semantic-aware prediction model is elaborated on a case of forecasting YouTube clip popularity.

In Multi-Agent Systems (MASs) it is of vital importance that basic network operations are performed without the interference of a central entity (i.e. agent). In this paper we will present how to use a self-organization approach to achieve time synchronization of agents in MASs using a Pulse-Coupled Oscillators (PCO) model that is based on flashing fireflies. Fireflies are known to emit flashes at regular intervals when isolated, but when they are within a group, they converge upon the same rhythm until time synchronization is achieved. This paper investigates how the choice of overlay network topology and metric affects the time synchronization process of agents in MASs.

This paper explores the convergence of Machine-to-Machine (M2M) networks with Network-Centric Operations (NCO). An overview of the M2M communication paradigm and current standardization efforts regarding architecture are given. It includes an overview of the network-centric networking and analyses the idea of implementing NCO approach to achieve situational awareness and self-synchronization of autonomous and intelligent M2M devices in networked M2M environments. Case study involving M2M e-Health scenario is given and discussed as a proof of concept for the proposed convergence.

Internet has become an important part of our everyday life, as it has created new opportunities for business and offered new ways of social gathering. A very critical topic concerning decision making by online transactions is trust issue. This paper presents a formal representation of a sample e-commerce environment with communicating agents, including their properties and assets. Further, we describe rules of an investment game that agents are playing. We propose two different scenarios. In the first one, agents’ investment decisions are ad hoc. In the second one, there is trust management system deployed that affects the agents’ decisions. Based on developed simulation tool, we investigate the proposed scenarios. We show that with introducing trust management system agents filter out bad agents and only good ones prosper. We also discuss other evident effects of trust management system on agents and their final investments values based on simulations results.

We propose a bounded model checking (BMC) method for the verification of multi-agent systems’ (MASs). The MASs are modelled by deontic interleaved interpreted systems, and specifications are expressed in the logic

$\sc{RTECTLKD}$

. The verification approach is based on the state of the art solutions to BMC, one of the mainstream approaches in verification of reactive systems. We test our results on a typical communication scenario: train controller problem with faults.

Autonomous systems are used nowadays in more and more sectors from vehicles to domestic robots. They can make decisions on their own or interact with humans, thus their robustness and safety are properties of crucial importance. Due to the adaptive and context-aware nature of these systems, the testing of such properties is especially challenging. In this paper, we propose a model-based testing approach to capture the context and requirements of such systems, to automatically generate test data representing complex situations, and to evaluate test traces and compute test coverage metrics.

The paper deals with symbolic approaches to bounded model checking (BMC) for Linear Time Temporal Logic extended with the epistemic component (LTLK), interpreted over Interleaved Interpreted Systems. Two translations of BMC for LTLK to SAT and to operations on BDDs are presented. The translations have been implemented, tested, and compared with each other as well as with another tool on several benchmarks for MAS. Our experimental results reveal advantages and disadvantages of SAT- versus BDD-based BMC for LTLK.

In this paper we model message exchange in a trusted multi-agent system (MAS). Modelling such MAS allows us to study and analyse agents’ behaviour. We apply our model in context where secure agent communication is paramount, i.e. within the context of data-sensitive and time-sensitive web-applications. As a result we aim to reduce design errors of such applications, analyse agents’ behaviour in secure and compromised web applications and consequently improve security, reliability and usability.

We use model checking to study the pattern of message exchange among agents. As a verification technique, model checking assists in finding flaws and simplifying the design of an application. We simulate a secure transaction between a client and a server. Our model is realistic; we augment it by introducing discrete time, web-session management, and authentication.

The BP-calculus is a formalism based on the

π

-calculus and encoded in WS-BPEL. The BP-calculus is intended to specificaly model and verify Service Oriented Applications. One important feature of SOA is the ability to compose services that may dynamically evolve along runtime. Dynamic reconfiguration of services increases their availability, but puts accordingly, heavy demands for validation, verification, and evaluation. In this paper we formally model and analyze dynamic reconfigurations and their requirements in BP-calculus and show how reconfigurable components can be modeled using handlers that are essential parts of WS-BPEL language.

We propose a learning method for efficient team formation by self-interested agents in task oriented domains. Service requests on computer networks have recently been rapidly increasing. To improve the performance of such systems, issues with effective team formation to do tasks has attracted our interest. The main feature of the proposed method is learning from two-sided viewpoints, i.e., team leaders who have the initiative to form teams or team members who work in one of the teams that are solicited. For this purpose, we introduce three parameters to agents so that they can select their roles of being a leader or a member, then an agent can anticipate which other agents should be selected as team members and which team it should join. Our experiments demonstrated that the numbers of tasks executed by successfully generated teams increased by approximately 17% compared with a conventional method.

This paper addresses the real-time control of multiple agents in the presence of disturbances and non-convex collision avoidance constraints. The goal is to guarantee the convergence towards a tight formation. A single optimal control problem is solved based on a prediction of the future evolution of the system and the resulting controller is implemented in a centralized way. At the supervision level, it is shown that the decision about which agents should take on what role in the desired tight formation is equivalent with a classical pairing (or task assignment) problem. Furthermore, the pairing is re-evaluated at each iteration. The proposed method exhibits effective performance validated through some illustrative examples.

This paper proposes an agent-based approach to customizing a remote conversation support system for cognitively handicapped people. This remote conversation support system is designed as a web application to assist with conversation over the Internet. The system is built from ‘gadgets,’ each of which implements a particular conversation support function. Since the need for conversation support varies from person to person, such a system must be customized to suit the requirements of the two users who conduct the conversation. The proposed approach introduces a user agent that corresponds to a human user. The negotiation protocol for selecting the necessary gadgets is defined based on the FIPA interaction protocol. This paper describes how the proposed protocol can be used to determine the gadgets to be used during run-time.

Nowadays, a modern company operates as a network of companies, not as a single company. These companies are called virtual enterprises, whose main purpose is the maximal fulfilment of costumer needs. Based on this, the paper presents the construction of agile supply chains and as an example it details the model, the algorithms and the applications of a modern electronic freight and warehouse exchange, which is a type of virtual enterprises.

Cooperation in the supply chain requires a lot of information exchange between different companies. In order to achieve appropriate data exchange open communication standards are needed which are commonly agreed and approved. In the field of logistics a number of standards already defined. The aim of the paper is to compare existing communication standards to the market needs and expectations. Authors present conclusions from pilot survey conducted within selected group of transport users and logistics service providers regarding information exchange by planning and execution of transport services.

The paper is addressed to economic problems for which many different models can be proposed. In such situation the ensemble approach is natural way to improve the final prediction results. In particular, we present the method for the prediction improvement with ensemble method based on the multivariate decompositions. As a method for model results decomposition we present the smooth component analysis. The resulting components are classified as destructive and removed, or as constructive and recomposed. The classification of the components is based on the theoretical analysis of

MSE

error measure. The robustness of the method is validated through practical experiment of energy load consumption in Poland.

The fours model designed to support competence management are presented: the model of knowledge processing in Open and Distance Learning, the motivation model, the model of developing a curriculum oriented on competence acquisition, the model of competence development in an intangible production network. All the models are integrated through common parameters and criterions and are adapted to Digital Economy context. In the paper the approach to models integration is proposed. The models are designed to be implemented in competences acquiring process for knowledge repository development in the distance learning network (DLN). The DLN’s aim is to ‘produce’ the competence required on the job market.

Good organization of reverse logistics channels allows company to explore the additional value from extended life cycle of products. In order to be able to effectively plan flow of returns in recovery network a big number of data is needed. Moreover the exchange of information between the cooperating companies must be done in time efficient manner. The aim of this paper is to present a framework for information management among companies within recovery network. The recovery network for electronic equipment will be used as an example for theoretical considerations. The key factors and obstacles for coordination of recovery network will be identified. The requirements for integration platform functionality will be presented.

The food supply chain is a very demanding and dynamic sector of economy. The goods that flow through the supply chain are exposed to damage. The authors have developed a model which optimizes the cost of transport in the network. It is based on graph theory, and, in particular, the Generalized Flow Problem has been used. The main contribution of this paper is the implementation of this model in a simulation environment. It has allowed to research how defect ratios influence the structure of an optimal configuration of a supply chain.

The paper investigates the possibility of multi-agent systems application in distributed computation illustrated by an example of generic distributed computation multi-agent system. The system is defined upon finite state machine architecture with agent internal and external behavior. For testing purpose, the model was implemented in Fuzzy Cognitive Map (FCM). FCM is qualitative modeling and simulation technique that is based on discrete distributed computation. Our proposed multi-agent system has shown expected characteristics so the conclusion is that the proposed generic model could be used as a foundation for building distributed computation multi-agent systems.

A model decomposition allows transforming a graph-based problem into a set of subproblems to be processed in parallel. Prepared model should guarantee good efficiency of communication and synchronization among agents managing a distributed representation of a system. In this article we present the new approach to such a decomposition, reducing required cooperation among agents. The comparison with replicated complementary graphs approach is also presented.

Recent growth of the embedded system industry, both in size and complexity, resulted in numerous advantages for the consumer equipment and a few problems for the system designers. One of the problems is the software licensing problem in complex embedded systems. The fact that today’s embedded systems are evolved into fairly large and complex systems presents the need to effectively manage the embedded software functionality, without the need to recompile and rewrite the firmware components. This problem can be avoided by using fully functional software accompanied by a licensing agent to enforce the license permissions granted to an end user. The licensing agent enforces the permissions stored in encrypted license file located on a Secure Digital memory data card. Another advantage is the fact that the permissions can be updated remotely using a license distribution agent located at the manufacturer’s server and uploaded through Ethernet or Internet. Licensing agents applied to various embedded systems have the ability to communicate with other licensing agents and license distribution agent in order to exchange and verify supplied licenses. This paper proposes the software licensing agent model and implements the licensing agent in the existing laboratory based embedded system.