This paper proposes a cooperative and distributed framework to evaluate and optimize task offloading in Mobile Edge Computing (MEC). Each agent, representing either a user device or an edge domain, autonomously interacts with others through trust-driven recommendations and cluster formation. The proposed algorithm exploits this information to iteratively increase — and asymptotically converge over time to — the configuration that maximizes the collective utility of edge servers and user devices, i.e., the Average Performance ( A P ), which corresponds to a Nash equilibrium where only reliable agents are rewarded. Two synthetic indicators are introduced to model the main aspects of MEC collaboration: the Quality of Experience ( Q o E ), representing the perceived user-side performance, and the Convenience ( C ), expressing the server-side efficiency and resource cost. Experimental validation, performed over a simulated MEC environment with up to 1000 agents, shows a rapid convergence (within 20 iterations), a stable equilibrium with A P ≈ 0 . 92 , and robustness to variations in the simulated values of agents’ reliability. The results demonstrate that the proposed distributed algorithm achieves efficient, self-organized coordination among heterogeneous edge entities while maintaining scalability and fairness.

The introduction of trust-based approaches in social scenarios modeled as multi-agent systems (MAS) has been recognized as a valid solution to improve the effectiveness of these communities. In fact, they make interactions taking place in social scenarios much fruitful as possible, limiting or even avoiding malicious or fraudulent behaviors, including collusion. This is also the case of multi-layered neural networks (NN), which can face limited, incomplete, misleading, controversial or noisy datasets, produced by untrustworthy agents. Many strategies to deal with malicious agents in social networks have been proposed in the literature. One of the most effective is represented by Eigentrust, often adopted as a benchmark. It can be seen as a variation of PageRank, an algorithm for determining result rankings used by search engines like Google. Moreover, Eigentrust can also be viewed as a linear neural network whose architecture is represented by the graph of Web pages. A major drawback of Eigentrust is that it uses some additional information about agents that can be a priori considered particularly trustworthy, rewarding them in terms of reputation, while the non pre-trusted agents are penalized. In this paper, we propose a different strategy to detect malicious agents which does not modify the real reputation values of the honest ones. We introduce a measure of effectiveness when computing reputation in presence of malicious agents. Moreover, we define a metric of error useful to quantitatively determine how much an algorithm for the identification of malicious agents modifies the reputation scores of the honest ones. We have performed an experimental campaign of mathematical simulations on a dynamic multi-agent environment. The obtained results show that our method is more effective than Eigentrust in determining reputation values, presenting an error which is about a thousand times lower than the error produced by Eigentrust on medium-sized social networks.

The blockchain (BC) paradigm is currently modifying all the major Internet of Things (IoT) application domains, giving the opportunity of constructing decentralized environments in which trustful and anonymous activities can be efficiently performed. BC proposes an approach which assures data saved on a distributed ledger (DL) would be continuously synchronized and in such a way the ledger will remain consistent. The DL has the responsibility to maintain a higher level of consistency. Given a connected network, the optimum neighbor selection (ONS) of paths can be obtained by finding the minimum spanning tree of the network. However, none of past approaches proposed in the literature to construct ONS considered the problem of having nodes with low levels of reliability in the network and even malicious or fraudulent nodes, that is a situation very common in an IoT environment. In this article, we propose an optimized BC ONS algorithm, called trust-based ONS (TONS), which allows the miners to communicate with a globally optimized selection of neighbors. The algorithm can also guarantee that these nodes are the most reliable miners. We also describe an experimental simulation campaign we have performed to evaluate the effectiveness and the efficiency of our approach.

The introduction of the IoT technology and its pervasive penetration in our daily life implies that IoT smart objects can often participate in social events, and in this case their paradigm of interaction is commonly denoted as social IoT (SIoT). To make reliable transactions into a SIoT scenario, in this article, we introduce a multi-agent SIoT architecture, which integrates a reputation system based on a clustering of the SOs. In our framework, when a SO looks for a resource and detects a reliable partner having that resource, then the two SOs can interact to make a transaction, and at the end of the transaction each of them provides a feedback about the partner to the local reputation system of the edge server, to update the SOs’ reputation scores. Moreover, each edge server sends these reputation scores to the cloud, which updates the associated reputation values of the SOs, deriving from experiences coming from all the edge domains. This architecture provides a given object, moving from an edge domain to another one, with the possibility to have an updated value of its reputation, represented by the value stored in the cloud. We have validated our approach by a campaign of simulations, whose results seem particularly promising.

Abstract Routing in the Internet of Things (IoT) renders the protection against various network attacks as any attacker intrudes the routing mechanism for establishing the destructive mechanisms against the network, which insists the essentiality of the security protocols in IoT. Thus, the paper proposes a secure protocol based on an optimization algorithm, Monarch-Earthworm Algorithm (Monarch-EWA), which is the modification of the Monarch Butterfly algorithm using the Earthworm Optimization Algorithm (EWA) in order to render effective security to the network. Initially, the effective nodes are selected using the Deep Convolutional Neural Network (deep CNN) classifier based on the factors, trust and energy of the node, and stochastic gradient descent algorithm trains the deep CNN classifier. The secure nodes are involved in routing for which the secure multipath is chosen optimally using the proposed Monarch-EWA, which chooses the secure multipath based on the factors, energy and trust. The analysis of the proposed method in the presence of attacks, such as black hole, message replicate and distributed denial of service, reveals that the proposed method outperformed the existing methods. The proposed Monarch-EWA protocol acquired the maximal energy, throughput and detection rate of 0.2268 J, 48.2759% and 82.6231%, respectively, with the minimal delay of 0.0959 ms.

In Smart Factories, automated guided vehicles (AGVs) accomplish heterogeneous tasks as moving objects, restoring connectivity, or performing different manufacturing activities into production lines. These kinds of devices combine several capabilities, as artificial intelligence (visual and speech recognition, automatic fault detecting, proactive behavior) and mobility, into the so-called “mobile intelligence.” A typical scenario is represented by a workshop with a large number of mobile intelligent devices with associated agents, mutually interacting on their behalf. Here, to reach a given target by contemporary satisfying some basic requirements like effectiveness and efficiency, it is often necessary to organize ad hoc teams of free-moving vehicles, sensors, and smart devices. Therefore, a specific issue is the adequate representation of the reciprocal agent/device trustworthiness for advantaging such team formation processes within a smart factory environment. To this end, in this article, first, we define a trust measure based on reliability and reputation of AGVs, which are computed based on the feedbacks released for the AGVs activities in the factory; second, we design a trust framework exploiting the defined measures to support the formation of virtual, temporary, and trust-based teams of mobile intelligent devices; and third, we present a set of experimental results highlighting that the proposed trust framework can improve the workshop performance in terms of effectiveness and efficiency.

The “Internet of Things” (IoT) is requiring increasing computational capabilities which can be obtained from Machine-to-Machine cooperation among smart objects. The convergence of IoT and Multi-Agent Systems gives the opportunity to benefit of their social attitude to interact and cooperate for services. In a mobile and federated context, selecting reliable partners for cooperation is a complex task because the most part of devices are not referenced with respect to their trustworthiness. The aggregation of agents in groups, in each IoT environment, on the basis of reputation information can provide a solution to this problem. In order to spread device/agent reputation information in our distributed scenario without using a central repository, we propose to rely on the blockchain technology to support a reputation approach to form groups. In particular, information about the reputation capital, certified by the blockchain, is used by an algorithm to drive group formation processes of agents in IoT environments.

Graph robustness-the ability of a graph to preserve its connectivity after the loss of nodes and edges-has been extensively studied to quantify how social, biological, physical, and technical systems withstand to external damages. In this paper, we prove that graph robustness can be quickly estimated through the Randic index, a parameter introduced in chemistry to study organic compounds. We prove that Erdos-Renyj (ER) graphs are a good specimen of robust graphs because they lack of a clear modular structure; we derive an analytical expression for the Randic index of ER graphs and use ER graphs as an effective term of comparison to decide about graph robustness. Experiments on real datasets from different domains (scientific collaboration networks, content-sharing systems, co-purchase networks from an e-commerce platform, and a road network) show that real-life large graphs are more robust than ER ones with the same number of nodes and edges. We also observe that if node degree distribution closely follows a power law, then few edges contribute for more than half of the Randic index, thus indicating that the selective removal of those edges has devastating impact on graph robustness. Finally, we describe sampling-based algorithms to efficiently but accurately approximate the Randic index.

This track started in 2009 with opening remarks from the Chair observing that the computing cloud evolution depends on research efforts from the infrastructure providers creating next generation hardware that is service friendly, service developers that embed business service intelligence in the computing infrastructure to create distributed business workflow execution services and service providers who assure service delivery on a massive scale with global interoperability. The state of the art architecture and evolution of the cloud at that time was already increasing datacenter complexity by piling up new layers of management over the many layers that already exist. Since then, the scale of distributed applications and their management have taken a new dimension demanding tolerance to wild fluctuations both in workloads and available computing resource pools. There are many calls to go cloud native and architect applications using the many services provided by the cloud providers such as Amazon Web Services. On the other hand, there are also calls for avoiding vendor lock-in by going multi-cloud and becoming cloud agnostic. In this conference there is a new paper that proposes cloud agnostic approach with globally interoperable cloud network using private or public network while reducing the complexity of Virtual machine image motion across clouds. In addition, there are 7 papers describing advances in current distributed and cloud computing practices dealing with quality of service, adaptive algorithms and software defined network architectures.

Matching users profile in virtual communities can represent the most natural way of representing group homogeneity, i.e. how much the group members are mutually linked. However, optimizing profile matching does not guarantee the group cohesion, i.e. that the group will continue to be homogeneous in time. Moreover, computing profile matching in large virtual communities can be very expensive, and cannot be integrated in a fully distributed system. In the past, we have demonstrated that using users mutual trust, along with profile matching, can help to improve groups homogeneity. In this work we demonstrate, by an extended set of experiments on datasets extracted from real communities, that trust measures can effectively replace profile matching in order to optimize group’s cohesion. A further interesting result is represented by the fact that it is also possible to replace the global trust measure with a local measure of trust, called local reputation, which is not highly sensitive to the size of the network, thus allowing to perform computations which are limited on the size of the ego-network of the single node.

Detecting communities in graphs is a fundamental tool to understand the structure of Web-based systems and predict their evolution. Many community detection algorithms are designed to processundirected graphs(i.e., graphs with bidirectional edges) but many graphs on the Web-e.g., microblogging Web sites, trust networks or the Web graph itself-are oftendirected. Few community detection algorithms deal with directed graphs but we lack their experimental comparison. In this paper we evaluated some community detection algorithms across accuracy and scalability. A first group of algorithms (Label Propagation and Infomap) are explicitly designed to manage directed graphs while a second group (e.g., WalkTrap) simply ignores edge directionality; finally, a third group of algorithms (e.g., Eigenvector) maps input graphs onto undirected ones and extracts communities from the symmetrized version of the input graph. We ran our tests on both artificial and real graphs and, on artificial graphs, WalkTrap achieved the highest accuracy, closely followed by other algorithms; Label Propagation has outstanding performance in scalability on both artificial and real graphs. The Infomap algorithm showcased the best trade-off between accuracy and computational performance and, therefore, it has to be considered as a promising tool for Web Data Analytics purposes.

In collaborative Web-based platforms, user reputation scores are generally computed according to two orthogonal perspectives: (a) helpfulness-based reputation (HBR) scores and (b) centrality-based reputation (CBR) scores. In HBR approaches, the most reputable users are those who post the most helpful reviews according to the opinion of the members of their community. In CBR approaches, a “who-trusts-whom” network—known as a trust network —is available and the most reputable users occupy the most central position in the trust network, according to some definition of centrality. The identification of users featuring large HBR scores is one of the most important research issue in the field of Social Networks, and it is a critical success factor of many Web-based platforms like e-marketplaces, product review Web sites, and question-and-answering systems. Unfortunately, user reviews/ratings are often sparse, and this makes the calculation of HBR scores inaccurate. In contrast, CBR scores are relatively easy to calculate provided that the topology of the trust network is known. In this article, we investigate if CBR scores are effective to predict HBR ones, and, to perform our study, we used real-life datasets extracted from CIAO and Epinions (two product review Web sites) and Wikipedia and applied five popular centrality measures—Degree Centrality, Closeness Centrality, Betweenness Centrality, PageRank and Eigenvector Centrality—to calculate CBR scores. Our analysis provides a positive answer to our research question: CBR scores allow for predicting HBR ones and Eigenvector Centrality was found to be the most important predictor. Our findings prove that we can leverage trust relationships to spot those users producing the most helpful reviews for the whole community.

Summary Recently, an ever increasing number of e‐Commerce tools has been made available that are able to help customers by generating purposed recommendations. Many of them are centralized so that they have to face problems related to efficiency and scalability. A few of them are distributed, but in this case, the complexity of the e‐Commerce process implies computation overhead on the client side, which is often unsuitable if mobile devices are used by customers. In this paper, we study how the software distribution in recommender systems affects their performances, depending on the characteristics of the e‐Commerce population. To this end, we present a distributed testbed architecture for e‐Commerce recommender systems using a multi‐tiered agent‐based approach to generate effective recommendations without requiring such an onerous amount of computation per single client. We use such a testbed to study the main advantages and limitations associated with the problem of distributing the computation of recommendations. Copyright © 2016 John Wiley & Sons, Ltd.

In this work, we propose an algorithm, called E-Learning Ontology Enrichment (ELOE), to derive a global representation from the personal ontologies of different agents present in an e-Learning MAS. Using ELOE, each agent of a MAS-based e-Learning system can autonomously enrich its own ontology by using semantic negotiation and, at the same time, it can access to the global ontology to have a view of the terms used by all the other agents. Each term of the global ontology is associated with a set of meanings, and each meaning is associated with the set of agents that have used it. This way an agent, having to send a message to an interlocutor, is able of choosing from the global ontology the most suitable term with the most appropriate meaning. Only if the agent does not find in the global ontology the appropriate term, it will use a personal term that probably will lead to a new semantic negotiation process. This way, the use of the onerous task of the semantic negotiation will be limited to only the strictly necessary situations, and consequently the whole communication cost is decreased.

Thematic groups are gaining a lot of attention and high centrality in OSNs, as users share opinions and/or mutually collaborate for reaching their targets. Since users can be affiliated with groups belonging to different social networks, they can be supported by personal software agents able to perform activities aimed at supporting relationships and mutual cooperation among heterogeneous OSNs thematic groups. Basically software agents can encode users profiles with detailed information to be related with specific groups. This work discusses a multi-agent framework whose structure focuses on the role of software agents and the use of a common, shared dictionary for each group. Personal software agents are associated to OSN users to share knowledge for a specific theme for a number of groups related to that topic. Group agents are defined to support each group on each OSN by interacting with personal agents to manage group affiliation and enrich the common dictionaries of their own groups. The common dictionary of the group is a key element to provide knowledge sharing and interoperability between personal and group agents. In the proposed approach each user agent is able to personalize its own dictionary and enrich that of its own groups by means of selected categories.

Learning software agents are able to assist Cloud providers in taking decisions about resource management at any level, as they are able to collect knowledge and improve their performances over time by means of learning strategies. On the other hand Cloud Federations allow providers to share computational infrastructures in order to build a distributed, interoperable multi-cloud context. In this work we present an evolutionary approach based on agent cloning, i.e. a mechanism of agent reproduction allowing providers to substitute an "unsatisfactory" agent acting in a "cloud context" with a clone of an existing agent having a suitable knowledge and a good reputation in the multi-cloud context. By this approach, cloud agents performances can be improved because they are substituted with agent clones that have shown a better behaviour.

Cloud Computing is a versatile computing paradigm capable of attracting a wide variety of applications. However, the necessity to provide a wide range of complex services led providers to establish mutual agreements to provide large-scale distributed multi-cloud environments. Providers gain the opportunity to compose service workflows that are effective and efficient, taking resources of their own competitors, and the capability to satisfy unexpected workload peaks. In this paper, we propose a reputation-based model aiming at supporting the service composition by considering measures of QoS collected by the measuring systems, and reputation measures collected with the customers by means of users feedback.

Summary In this paper, a distributed approach aimed at improving the quality of service in dynamic grid federations is presented. Virtual organizations (VO) are grouped into large‐scale federations in which the original goals and scheduling mechanisms are left unchanged, while grid nodes can be quickly instructed to join or leave any VO at any time. Moreover, an agent‐oriented framework is designed to observe and characterize past behaviors of nodes in terms of resource sharing and consumption, as well as to determine the trust relationships occurring between each pair of nodes. By combining trust and historical behaviors into a unified convenience measure, software agents are able to evaluate the (i) advantages of node's membership with VOs and (ii) whether a specific set of nodes is able to meet the actual requirements, in terms of resource sharing and consumption of a specific VO. The convenience measure has been exploited to design a fully decentralized, greedy procedure, aimed at controlling the grid formation process. Extensive simulations have shown that the coordinated and decentralized process of grid formation provides a powerful means to improve the overall quality of service of the grid federation. Copyright © 2015 John Wiley & Sons, Ltd.

Abstract Most of the activities usually performed by Web users are today effectively supported by using appropriate metadata that make the Web practically readable by software agents operating as users’ assistants. While the original use of metadata mostly focused on improving queries on Web knowledge bases, as in the case of SPARQL-based applications on RDF data, other approaches have been proposed to exploit the semantic information contained in metadata for performing more sophisticated knowledge discovery tasks. Finding semantic associations between Web data seems a promising framework in this context, since it allows that novel, potentially interesting information can emerge by the Web’s sea, deeply exploiting the semantic relationships represented by metadata. However, the approaches for finding semantic associations proposed in the past do not seem to consider how Web entities are logically collected into groups, that often have a complex hierarchical structure. In this paper, we focus on the importance of taking into account this additional information, and we propose an approach for finding semantic associations which would not emerge without considering the structure of the data groups. Our approach is based on the introduction of a new metadata model, that is an extension of the direct, labelled graph allowing the possibility to have nodes with a hierarchical structure. To evaluate our approach, we have implemented it on the top of an existing recommender system for Web users, experimentally analyzing the introduced advantages in terms of effectiveness of the recommendation activity.

Understanding the dynamics behind group formation and evolution in social networks is considered an instrumental milestone to better describe how individuals gather and form communities, how they enjoy and share the platform contents, how they are driven by their preferences/tastes, and how their behaviors are influenced by peers. In this context, the notion of compactness of a social group is particularly relevant. While the literature usually refers to compactness as a measure to merely determine how much members of a group are similar among each other, we argue that the mutual trustworthiness between the members should be considered as an important factor in defining such a term. In fact, trust has profound effects on the dynamics of group formation and their evolution: individuals are more likely to join with and stay in a group if they can trust other group members. In this paper, we propose a quantitative measure of group compactness that takes into account both the similarity and the trustworthiness among users, and we present an algorithm to optimize such a measure. We provide empirical results, obtained from the real social networks EPINIONS and CIAO, that compare our notion of compactness versus the traditional notion of user similarity, clearly proving the advantages of our approach.

Cloud Computing is the reference model for IT specialists offering high quality services in a pay-per-use fashion. Architectures and protocols for Cloud Federations have been designed to meet emerging requirements of Cloud infrastructure. In such a context IT providers can mutually rent virtual resources and exchange metadata in order to get more opportunity. In this paper we define VST - Virtual machine Software Tracking - as the process of extracting detailed information about software running in virtual machines in a transparent manner. Coupling VST data with those of resource consumption can help IaaS providers to be more precise when assigning resources to VMs. Moreover, sharing VST information across the cloud federation can help IaaS providers to get decision about VM placement in time. For this aim we provide an agent based architecture for collecting and sharing VST data in a Cloud Federation, together with a skeleton implementation by which the architecture has been tested.

This paper proposes a trust model for a competitive federation of Cloud computing systems. The reference scenario is a large-scale environment in which different Cloud providers compete in offering their pay-per-use XaaS (Anything as a Service) resources to the various users. In this context, to obtain the assignment of a user request, providers may state to possess more resources than the amount really available: the result could be the provisioning of a services with a lower QoS than expected. A trust model is thus mandatory to help the requester to properly choose the adequate provider on the basis of (i) the reliability shown in the past in providing services, and (ii) the reputation the provider has to other users. In this sense, the paper presents an extension to the RRAF model, which has been suitably adapted to a large-scale environment by means of the introduction of a multi-dimensional approach which fits the various reliability issues, typical of Cloud computing systems.

A main issue in competitive Multi-Agent Systems is that of allowing self-interested agents to mutually cooperate in those cases where the local agent resources are not sufficient to satisfy user requests.To this end, it is necessary to introduce an internal organization for allowing agents to form suitable friendships and groups facilitating the collaboration.In the past, several approaches have been proposed aiming at forming agent coalitions which maximize the profit of the group or the individual agent.However, this viewpoint could introduce some negative side-effects, namely (i) it can lead to reward the most aggressive agents, also if they have bad social behaviours or (ii) it could introduce a sort of social flattening, without taking into account the differences among the agents in terms of merit.To face this issue, in this paper we propose an algorithm for forming friendships and groups which, instead of maximizing individual or global profit, tries to optimize a social capital represented by the mutual trust relationships.We theoretically prove that the application of our algorithm leads the competition to reward those agents exhibiting the most virtuous behaviours, introducing meritocracy in the system, not only rewarding effective agent performances but also encouraging correct behaviours.

In an agent-based cloud scenario, when two agents, provided with different ontologies exchange messages to negotiate a Service Level Agreement for a cloud service, they could fail to understand both the correct meaning of their messages and the offered service quality, due to the differences present in their ontologies. In all these cases, a semantic negotiation is required to reciprocally making understandable the messages exchanged about the cloud service characteristics. In this work we propose a protocol for supporting the SLA negotiation process, involving semantic issues, which do not require the use of a common global agent ontology sharing knowledge in advance. In detail, this protocol considers that each agent is able to partition the other agents based on both their expertise in cloud services and similarity with their own ontologies.

Several models have been proposed in the past for representing both reliability and reputation. However, we remark that a crucial point in the practical use of these two measures is represented by the possibility of suitably combining them to support the agent's decision. In the past, we proposed a reliability–reputation model, called RRAF, that allows the user to choose how much importance to give to the reliability with respect to the reputation. However, RRAF shows some limitations, namely: (i) The weight to assign to the reliability versus reputation is arbitrarily set by the user, without considering the system evolution; (ii) the trust measure that an agent a perceives about an agent b is completely independent of the trust measure perceived by each other agent c, while in the reality the trust measures are mutually dependent. In this paper, we propose an extension of RRAF, aiming at facing the limitations above. In particular, we introduce a new trust reputation model, called TRR, that considers, from a mathematical viewpoint, the interdependence among all the trust measures computed in the systems. Moreover, this model dynamically computes a parameter measuring the importance of the reliability with respect to the reputation. Some experiments performed on the well-known ART(Agent Reputation and Trust) platform show the significant advantages in terms of effectiveness introduced by TRR with respect to RRAF. © 2011 Wiley Periodicals, Inc.

Ontologies significantly enhance the possibility to make agent-based Web applications really “semantic”. However, in order to make their usage both effective and efficient, some alignment problems, arising from the presence of semantic heterogeneitie

Personalization is becoming a key issue in designing effective e‐learning systems and, in this context, a promising solution is represented by software agents. Usually, these systems provide the student with a student agent that interacts with a site agent associated with each e‐learning site. However, in presence of a large number of students and of e‐learning sites, the tasks of the agents are often onerous, even more if the student agents run on devices with limited resources. To face this problem, we propose a new multiagent learning system, called ISABEL. Our system provides each student, that are using a specific device, with a device agent able to autonomously monitor the student's behavior when accessing e‐learning Web sites. Each site is associated, in its turn, with a teacher agent. When a student visits an e‐learning site, the teacher agent collaborates with some tutor agents associated with the student, to provide him with useful recommendations. We present both theoretical and experimental results to show that this distributed approach introduces significant advantages in quality and efficiency of the recommendation activity with respect to the performances of other past recommenders.

Social internetworking systems are a significantly emerging new reality; they group together a set of social networks and allow their users to share resources, to acquire opinions and, more in general, to interact, even if these users belong to different social networks and, therefore, did not previously know each other. In this context the notions of trust and reputation play a very relevant role. These notions have been widely studied in the past in several contexts whereas they have been largely neglected in the social internetworking research; however, since this application field presents several peculiarities, the results found in other application contexts are not automatically valid here. This paper introduces a model to represent and handle trust and reputation in a social internetworking system and proposes an approach that exploits these parameters to provide users with suggestions about the most reliable persons they can contact or social networks they can register to.

Web recommender systems are Web applications capable of generating useful suggestions for visitors of Internet sites. However, in the case of large user communities and in presence of a high number of Web sites, these tasks are computationally onerous, even more if the client software runs on devices with limited resources. Moreover, the quality of the recommendations strictly depends on how the recommendation algorithm takes into account the currently used device. Some approaches proposed in the literature provide multidimensional recommendations considering, besides items and users, also the exploited device. However, these systems do not efficiently perform, since they assign to either the client or the server the arduous cost of computing recommendations. In this article, we argue that a fully distributed organization is a suitable solution to improve the efficiency of multidimensional recommender systems. In order to address these issues, we propose a novel distributed architecture, called MUADDIB, where each user's device is provided with a device assistant that autonomously retrieves information about the user's behavior. Moreover, a single profiler , associated with the user, periodically collects information coming from the different user's device assistants to construct a global user's profile. In order to generate recommendations, a recommender precomputes data provided by the profilers. This way, the site manager has only the task of suitably presenting the content of the site, while the computation of the recommendations is assigned to the other distributed components. Some experiments conducted on real data and using some well-known metrics show that the system works more effectively and efficiently than other device-based distributed recommenders.

Forming groups of agents is an important task in many agent-based applications, for example when determining a coalition of buyers in an e-commerce community or organizing different Web services in a Web services' composition. A key issue in this context is that of generating groups of agents such that the communication among agents of the same group is not subjected to comprehension problems. To this purpose, several approaches have been proposed in the past in order to form groups of agents based on some similarity measures among agents. Such similarity measures are mainly based on lexical and/or structural similarities among agent ontologies. However, the necessity of taking into account a semantic component of the similarity value arises, for example by considering the context in which a term is used in an agent ontology. Therefore we propose a clustering technique based on the HISENE semantic negotiation protocol, using a similarity value that has lexical, structural and semantic components. Moreover, we introduce a suitable multiagent architecture that allows computing agent similarities by means of an efficient distributed approach.

Agent-based e-learning systems allow the interaction between students and e-learning Web sites, providing students with useful suggestions about the available educational resources. In these systems, generally each student is monitored by a student agent, while each e-learning site is associated with a site agent. However, these systems are not able to support students which exploit different devices as PC, palmtop, cellular, etc. multi-agent system handling adaptivity (MASHA) is a recent multi-agent system which appears as a promising candidate to overcome such a limitation. However, in the case of large agent communities, the tasks of both the student and the site agent in MASHA can result significantly heavy. To face this issue, we propose in this paper an extension of the MASHA system, called MASHA-EL, appositely conceived for supporting e-learning. In this system, a student that exploits a given device is provided with a device agent, and each e-learning Web site is associated, in its turn, with a teacher agent. When a student visits the e-Learning site of the teacher agent using a given device, the teacher agent provide him with useful recommendations also considering the exploited device. We present some experimental results that compares the performances of MASHA-EL with other well- known agent-based recommenders, and that show significant advantages obtained by MASHA-EL in terms of recommendation effectiveness and reduction of time costs.

In this article, we propose an agent-based approach for managing e-commerce activities. In our approach, an agent is present in each e-commerce site, managing the information stored there. In addition, another agent is associated with each customer, handling his/her profile. The proposed approach is based on the use of a particular conceptual model called the Behaviour-Semantic Distance and Relevance (B-SDR) network, which is capable of uniformly representing and handling information stored in e-commerce sites and customer profiles. The capabilities of the B-SDR network model are exploited to let customer and site agents cooperate in such a way in order to support a customer in identifying, whenever he/she accesses an e-commerce site, those products and services present in the site itself and for better matching his/her interests. The approach has been implemented in a prototype in which its functionalities are discussed here also. © 2004 Wiley Periodicals, Inc.

Histograms are used to summarize the contents of relations for the estimation of query result sizes into a number of buckets. Several techniques (e.g., MaxDiff and V-Optimal) have been proposed in the past for determining bucket boundaries which provide better estimations. This paper proposes to use 32 bit information (4-level tree index) for each bucket for storing approximated cumulative frequencies at 7 internal intervals of a bucket. Both theoretical analysis and experimental results show that the 4-level tree index provides the best frequency estimation inside a bucket. The index is later added to two well-known techniques for constructing histograms, MaxDiff and V-Optimal, thus obtaining high improvements in the frequency estimation over inter-bucket ranges w.r.t. the original methods.

In this paper we propose a multi-agent model for handling e-commerce activities. In our model, an agent is present in each e-commerce site, managing the information stored therein. In addition, another agent is associated with each customer handling her/his profile. The proposed model is based on the exploitation of a particular conceptual model, called the B-SDR network, capable of representing and handling both information stored in e-commerce sites and customer profiles. The capabilities of the B-SDR network model are exploited to let customer and site agents to cooperate in such a way to support a customer to detect, whenever she/he accesses an e-commerce site, those products and services present in the site itself and better matching her/his interests.