Skip to main content

Über dieses Buch

This book constitutes thoroughly revised, selected papers of the proceedings of the 6th International Workshop on Collaborative Agents Research and Development, CARE 2015 and the Second International Workshop on Multi-agent Foundations of Social Computing, MFSC 2015, held in Istanbul, Turkey, on May 4, 2015. Both Workshops were held in conjunction with AAMAS 2015.

The 5 revised full papers of CARE and the 7 full papers of MFSC presented were carefully selected from 14 CARE and 10 MFSC submissions. Both workshop address issues in relevant areas of social computing such as smart societies, social applications, urban intelligence, intelligent mobile services, models of teamwork and collaboration.



Automated Negotiation for Traffic Regulation

Urban congestion is a major problem in our society for quality of life and for productivity. The increasing communication abilities of vehicles and recent advances in artificial intelligence allow new solutions to be considered for traffic regulation, based on real-time information and distributed cooperative decision-making models. The paper presents a mechanism allowing a distributed regulation of the right-of-way of the vehicles at an intersection. The decision-making relies on an automatic negotiation between vehicles equipped with communication devices, taking into account the travel context and the constraints of each vehicle. During this negotiation, the vehicles exchange arguments, in order to take into account various types of information, on individual and network scales. Our mechanism deals with the continuous aspect of the traffic flow and performs a real-time regulation.
Matthis Gaciarz, Samir Aknine, Neila Bhouri

Towards a Middleware for Context-Aware Health Monitoring

The surge of commodity devices, sensors and apps allows for the continuous monitoring of patient’s health status with relatively low-cost technology. Nonetheless, current solutions focus on presenting data and target at individual health metrics and not intelligent recommendations. In order to advance the state-of-the-art, there is a demand for models that correlate mobile sensor data, health parameters, and situational and/or social environment. We seek to improve current models by combining environmental monitoring, personal data collecting, and predictive analytics. For that, we introduce a middleware called Device Nimbus that provides the structures to integrate data from sensors in existing mobile computing technology. Moreover, it includes the algorithms for context inference and recommendation support. This development leads to innovative solutions in continuous health monitoring, based on recommendations contextualised in the situation and social environment. In this paper we propose a model, position it against state-of-the-art, and outline a proof-of-concept implementation.
Eduardo A. Oliveira, Fernando Koch, Michael Kirley, Carlos Victor G. dos Passos Barros

The Influence of Users’ Personality on the Perception of Intelligent Virtual Agents’ Personality and the Trust Within a Collaborative Context

As Intelligent Virtual Agents (IVAs) have been widely used for applications that require human interaction and collaboration, modeling an IVA that can exhibit personalities is becoming increasingly important. A large body of research has studied variant verbal and non-verbal aspects that are used to deduce an IVA’s personality; however, research falls short in showing whether humans’ personality influences their perception of the IVA’s personality. This paper presents an empirical study that investigated whether human users can perceive the intended personality of an IVA through verbal and/or non-verbal communication, on one hand, and the influence of the user’s own personality on their perception, on the other hand. Furthermore, we investigated whether the perceived personality had an impact on the human’s level of trust in the IVA teammate. The results showed that similarity in personalities between humans and IVAs tended to significantly influence the humans’ correct perception of the IVA’s personality and that different perceived personalities influenced the human’s level of trust.
Nader Hanna, Deborah Richards

The Effects of Temperament and Team Formation Mechanism on Collaborative Learning of Knowledge and Skill in Short-Term Projects

While collaborative learning has long been believed to hold a great value for organizations and classrooms, Modeling this learning in small, short-term project teams is a challenge. This paper describes the development of an agent-based modeling approach that can assist in understanding the collaborative learning of such project teams. A key aspect of the presented approach is our distinction between knowledge and skills required for the achievement of project goals. Both of these forms of intelligence need to be learned in the project context, but the rate of their expansion or enhancement may proceed differently, depending on the personality makeup of the team and the mechanism employed for team assembly. Based on reports from the theoretical and empirical literature, we derive a multi-agent computational model that characterizes how knowledge and skills may be learned among team members with varying personality attributes. Also, Group formation in virtual learning environments is either done voluntary or with the support from the system. In this connection, we studied two types of group formation mechanisms and the role of each mechanism in the collaborative learning and performance of teams.
Mehdi Farhangian, Martin Purvis, Maryam Purvis, Tony Bastin Roy Savarimuthu

Exploring Smart Environments Through Human Computation for Enhancing Blind Navigation

In this paper the orchestration of wearable sensors with human computation is explored to provide map metadata for blind navigation. Technological navigation aids for blind must provide accurate information about the environment and select the best path to reach a chosen destination. Urban barriers represent dangers for the blind users. The dynamism of smart cities promotes a constant change of these dangers and therefore a potentially “dangerous territory” for these users. Previous work demonstrated that redundant solutions in smart environments complemented by human computation could provide a reliable and trustful data source for a new generation of blind navigation systems. We propose and discuss a modular architecture, which interacts with environmental sensors to gather information and process the acquired data with advanced algorithms empowered by human computation. The gathered metadata should enable the creation of “happy maps” that are delivered to blind users through a previously developed navigation system.
Hugo Paredes, Hugo Fernandes, André Sousa, Luis Fernandes, Fernando Koch, Renata Fortes, Vitor Filipe, João Barroso

Incorporating Mitigating Circumstances into Reputation Assessment

Reputation enables customers to select between providers, and balance risk against other aspects of service provision. For new providers that have yet to establish a track record, negative ratings can significantly impact on their chances of being selected. Existing work has shown that malicious or inaccurate reviews, and subjective differences, can be accounted for. However, an honest balanced review of service provision may still be an unreliable predictor of future performance if the circumstances differ. Specifically, mitigating circumstances may have affected previous provision. For example, while a delivery service may generally be reliable, a particular delivery may be delayed by unexpected flooding. A common way to ameliorate such effects is by weighting the influence of past events on reputation by their recency. In this paper, we argue that it is more effective to query detailed records of service provision, using patterns that describe the circumstances to determine the significance of previous interactions.
Simon Miles, Nathan Griffiths

Agent Protocols for Social Computation

Despite the fact that social computation systems involve interaction mechanisms that closely resemble well-known models of agent coordination, current applications in this area make little or no use of the techniques the agent-based systems literature has to offer. In order to bridge this gap, this paper proposes a data-driven method for defining and deploying agent interaction protocols that is entirely based on using the standard architecture of the World Wide Web. This obviates the need of bespoke message passing mechanisms and agent platforms, thereby facilitating the use of agent coordination principles in standard Web-based applications. We describe a prototypical implementation of the architecture and experimental results that prove it can deliver the scalability and robustness required of modern social computation applications while maintaining the expressiveness and versatility of agent interaction protocols.
Michael Rovatsos, Dimitrios Diochnos, Matei Craciun

Negotiating Privacy Constraints in Online Social Networks

Privacy is a major concern of Web systems. Traditional Web systems employ static privacy agreements to notify its users of how their information will be used. Recent social networks allow users to specify some privacy concerns, thus providing a partially personalized privacy setting. However, still privacy violations are taking place because of different privacy concerns, based on context, audience, or content that cannot be enumerated by a user up front. Accordingly, we propose that privacy should be handled per post and on demand among all that might be affected. To realize this, we envision a multiagent system where each user in a social network is represented by an agent. When a user engages in an activity that could jeopardize a user’s privacy (e.g., publishing a picture), agents of the users negotiate on the privacy concerns that will govern the content. We employ a negotiation protocol and use it to settle differences in privacy expectations. We develop a novel agent that represents its user’s preferences semantically and reason on privacy concerns effectively. Execution of our agent on privacy scenarios from the literature show that our approach can handle and resolve realistic privacy violations before they occur.
Yavuz Mester, Nadin Kökciyan, Pınar Yolum

Agent-Based Modeling of Resource Allocation in Software Projects Based on Personality and Skill

The success or failure of software development group work depends on the group members’ personalities, as well as their skills in performing various tasks associated with the project. Moreover, in the reality, tasks have a dynamic nature and their requirements change over time. Therefore, the effect of task dynamics on the teamwork must be taken into consideration. To do so, after describing a general approach to select effective team members based on their personalities and skills, we consider as an example a comparative multi-agent simulation study contrasting two different sample strategies that managers could use to select team members: by minimizing team over-competency and by minimizing team under-competency. Based on the simulation results, we drive a set of propositions about the conditions under which there are and are not performance benefits from employing a particular strategy for task allocation. Also, we propose a simulation environment that could provide a low cost tool for managers and researchers to gain better insights about effectiveness of different task allocation strategies and employees with different attributes in dynamic environments.
Mehdi Farhangian, Martin Purvis, Maryam Purvis, Tony Bastin Roy Savarimuthu

On Formalizing Opportunism Based on Situation Calculus

In social interactions, it is common for one party to possess more or better knowledge about a specific transaction than others. In this situation, parties who are more knowledgeable might perform opportunistic behavior to others, which is against others’ interest thus leading to relationship deterioration. In this study, we propose formal models of opportunism, which consist of the properties knowledge asymmetry, value opposition and intention, based on situation calculus in different context settings. We illustrate our formalization through a simple example. Further study on its emergence and constraint mechanism can be carried out based on the formal models.
Jieting Luo, John-Jules Meyer, Frank Dignum

Programming JADE and Jason Agents Based on Social Relationships Using a Uniform Approach

Interaction is an essential feature in multiagent systems. Design primitives are needed to explicitly model desired patterns. This work presents 2COMM as a framework for defining social relations among parties, represented by social commitments. Starting from the definition of interaction protocols, 2COMM allows to decouple interaction design from agent design. Currently, adapters were developed for allowing the use of 2COMM with the JADE and the JaCaMo platforms. We show how agents for the two platforms can be implemented by relying on a common programming schema.
Matteo Baldoni, Cristina Baroglio, Federico Capuzzimati

The Emergence of Norms via Contextual Agreements in Open Societies

This paper explores the emergence of norms in agents’ societies when agents play multiple - even incompatible - roles in their social contexts simultaneously, and have limited interaction ranges. Specifically, this article proposes two reinforcement learning methods for agents to compute agreements on strategies for using common resources to perform joint tasks. The computation of norms by considering agents’ playing multiple roles in their social contexts has not been studied before. To make the problem even more realistic for open societies, we do not assume that agents share knowledge on their common resources. So, they have to compute semantic agreements towards performing their joint actions.
George A. Vouros


Weitere Informationen

Premium Partner