Workshop on Trust Technologies (WG5) in Cyprus

Start Date: 2009-12-15

End Date: 2009-12-16

Location: Ayia Napa

Country: Cyprus

Description:

The purpose of this working group meeting is to address two issues that were suggested as highly relevant by the attendees of the kick-off workshop hold in Budapest back in May. These two issues are the modelling of trust on things and trust on groups. Although it is rather common to think of trust as a relationship between commitments or agreements and actual behaviour, the concept of trust on things seems rather different as things don’t have a capability of commitment, and trust on groups seems also different as the notion of group commitment is a debatable notion. Any models, experiments, or philosophical position papers are welcome for discussion.

::

PARTICIPANTS SLIDES:

::

::

The workshop will be organised in three sessions, one for each topic and one on general trust issues. We’ll also hold joint panels with other WGs and participate in the Agreement Technologies community meeting.

The agenda will then be organised as follows:

Tuesday 15 December 2009
Wednesday 16 December 2009
9:30 - 11:00
General Topics
11:00 - 11:30
Coffee
11:30 - 13:00
Panel on Norms, Organisations, and Semantics
13:00 - 14:30
Lunch
Lunch
14:30 - 16:00
Network Issues
Panel on Trust, Argumentation and Semantics
16:00 - 16:30
Afternoon Tea
Afternoon Tea
16:30 - 18:00
Trust and Structure
Agreement Technologies Community Meeting

Any researcher interested in the topic is requested to send an expression of interest to Anna Enciso (anna-at-iiia.csic.es) as an email. The expression of interest should indicate whether the researcher is interested in attending the working group activities on the two days mentioned above and also whether he or she would like to give a talk on models, ideas, results, experiments related to the topics of the working group as expressed above. In this second case, please include in the expression of interest:

1) Title of the contribution

2) Abstract of 200 words

3) Amount of requested time slot for the presentation

We’ll try and accommodate all time slots according to the overall available time.

Please send your expressions of interest before November 15 so we can plan the program well in advance.

The workshop will take place at the Grecian Bay Hotel, Ayia Napa, Cyprus, Colocated with EUMAS-2009.

::

PROGRAMME:

15 December 2009:

14:30 - 16:00: Network Issues (20 minutes per talk)

  • Jose Miguel Such, Universitat Politècnica de València, Spain

Title: Trust and Reputation using Partial Identities

  • Ioanna Dionysiou, University of Nicosia, Cyprus

Title: Adaptive Trust in Grid Environments

  • Àngels Fàbregues, IIIA-CSIC, Spain

Title: A testbed for Multiagent Systems

  • Francesca Toni, Imperial College London, United Kingdom

Title: Argumentation for trust in service-oriented architectures

16:30 - 18:00: Trust and Structure (20 minutes per talk)

  • Andrew Koster, IIIA-CSIC, Spain

Title: Inductively Generated Trust Alignments based on Shared Interactions

  • Ramón Hermoso, University Rey Juan Carlos, Spain

Title: Role Evolution in Open Multi-Agent Systems as an Information Source for Trust

  • Carles Sierra, IIIA-CSIC, Spain

Title: Propagation and Aggregation of Opinions in Structural Graphs

:

16 December 2009:

9:30 - 11:00: General Topics (20 minutes per talk)

  • Joana Urbano, University of Porto, Portugal

Title: Using Contextual Information to Tune the Estimation of Trust

  • Denis Trcek, University of Ljubljana, Slovenia

Title: Qualitative judgment dynamics

  • Jonathan Ben-Naim, IRIT, France

Title: Preliminary Results on Reputation Systems: Favoring more balanced support (20 minutes)

  • Carles Sierra, IIIA-CSIC, Spain

Title: WG5’s future activities

11:30 - 13:00: Panel on Norms, Organisations, and Semantics (Moderator: Michael Luck, King’s College, UK)

14:30 - 16:00: Panel on Trust, Argumentation and Semantics (Moderator: Francesca Toni, Imperial College, UK)

16:30- 18:00: Agreement Technologies Community Meeting

::

PARTICIPANTS (provisional):

Jonathan Ben-Naim, IRIT, France

Title: Preliminary Results on Reputation Systems: Favoring more balanced support (20 minutes)

Abstract:

In many large multi-agent systems (for example e-commerce applications), the users have to choose the agents to interact with. But this choice is difficult, in particular because the agents are too numerous. For the same reason, it is impossible to call on a trusted arbiter that can judge every agent. However, the agents provide a lot of information about their peers (for example feedbacks from past transactions). Consequently, reputation systems have been developed to exploit such information in order to provide an object helping the users to make decisions about the agents. The present paper contains work in progress, we develop and begin to analyze, from an axiomatic point of view, new reputation systems. The information available about the agents is modeled by a graph, that is, the nodes represent the agents and an arrow means that the source agent supports the importance of the destination agent. The object constructed from this information is a total ranking of the agents. When comparing two agents, there are essentially two criteria to consider : the number of supporters and the importance of the supporters. The ranking methods we have developed are based on the following principle: for an agent, to reach a certain reputation level, it is necessary to pass a certain threshold both in the quantity of supporters and in their quality.

::

Ioanna Dionysiou, University of Nicosia, Cyprus

Title: Adaptive Trust in Grid Environments (15-20 minutes)

Abstract:

Even though trust plays a significant role during decision-making in open collaborative environments, still end-user trust mechanisms have not be deployed in the EGEE grid computing setting. For example, if trust mechanisms were in place, a grid user could have chosen the trustworthiest site from a pool of available sites to submit a job. Or a grid user could have specified its trust requirements in a parameterized job description. Sites, offering computational resources, could have being rated based on their reputation among grid users.

It could appear that a grid user may assess the risk of having a job executed on a grid site by making a single trust assessment regarding the site’s ability to execute the job correctly and reliably. However, this is not sufficient. The presence of intermediate entities that further process the job affects the quality of job. As a result, a grid user has to make trust assessments for all interacting trustee entities that collaboratively execute the task and combine them in order to derive an end-to-end trust assessment about the quality of the submitted job.

In this talk, I will present a preliminary investigation on the integration of the trust paradigm in the grid environment, with a special focus on achieving end-user trust in an intuitive and practical manner.

::

Àngels Fàbregues, IIIA-CSIC, Spain

Title: A testbed for Multiagent Systems (20 minutes)

Abstract:

There is a chronic lack of shared application domains to test the research models and agent architectures on areas like negotiation, argumentation, trust and reputation. In this paper we introduce such a friendly testbed that we used for all such purposes. The testbed is based on the Diplomacy Game due to its lack of random moves and because of the essential role that negotiation and the relationships between the players play in the game. The testbed may also profit from the existence of a community of bot developers and a large number of human players that would provide data for our experiments. We offer the infrastructure and make it freely available to the MAS community.

::

Ramón Hermoso, University Rey Juan Carlos, Spain

Title: Role Evolution in Open Multi-Agent Systems as an Information Source for Trust (20 minutes)

Abstract:

In Open Multi-Agent Systems (OMAS), deciding with whom to interact is a particularly difficult task for an agent, as repeated interactions with the same agents are scarce, and reputation mechanisms become increasingly unrealiable. In this work, we present a coordination artifact which can be used by agents in an OMAS to take more informed decisions regarding partner selection, and thus to improve their individual utilities. This artifact monitors the interactions in the OMAS, evolves a role taxonomy, and assigns agents to roles based on their observed performance in different types of interactions. This information can be used by agents to better estimate the expected behaviour of potential counterparts in future interactions. We thus highlight the descriptive features of roles, providing expectations of the behaviour of agents in certain types of interactions, rather than their normative facets. We empirically show that the use of the artifact helps agents to select better partners for their interactions than selection processes based only on agents’ own experience. This is especially significant for agents that are newcomers to the OMAS.

..

Andrew Koster, IIIA-CSIC, Spain

Title: Inductively Generated Trust Alignments based on Shared Interactions (20 minutes)

Abstract:

In open multi-agent systems trust models are an important tool for agents to achieve effective interactions. However, in these kinds of open systems, the agents do not necessarily use the same, or even similar, trust models, leading to semantic differences between trust evaluations in the different agents. Hence, to successfully use communicated trust evaluations, the agents need to align their trust models.
We explicate that currently proposed solutions, such as common ontologies or ontology alignment methods, lead to additional problems and propose a novel approach. We show how the trust alignment can be formed by considering the interactions agents share. We describe a method, which uses inductive learning algorithms, to accomplish this alignment and test it in an example scenario.

::

Carles Sierra, IIIA-CSIC, Spain

Title: Propagation and Aggregation of Opinions in Structural Graphs

Abstract:

Trust and reputation measures are crucial in distributed open systems where agents need to decide whom or what to choose. Existing work has mainly focused on the reputation of single entities, neglecting their position amongst others and its effect on the propagation of trust. This paper presents an algorithm for the propagation of reputation in structural graphs. It focuses on the “part of” relation to illustrate how reputation may flow (or propagate) from one entity to another. The paper bases its reputation measures on opinions, which it defines as probability distributions over an evaluation space. Such a definition allows for a richer representation of opinions and leads towards new aggregation mechanisms.

::f

Jose Miguel Such, Universitat Politècnica de València, Spain

Title: Trust and Reputation using Partial Identities (15-20 minutes)

Abstract:

Currently, there are two different approaches to information system security: hard security, involving traditional security mechanisms like authentication, authorization, etc.; and soft security, involving social control mechanisms in general. Both approaches to security are complementary and can be mixed in order to achieve greater levels of security. What is more, when relationships between these two approaches are not taken into account, some vulnerabilities can emerge which otherwise would not. Therefore, we strongly encourage research transversal to these two approaches to security.
This study explores the relationships between the hard security concepts of identity and privacy on the one hand, and the soft security concepts of trust and reputation on the other hand. We propose definitions of identity and related concepts to be taken into account by trust and reputation systems. We also show how identities can help to define the context in which trust and reputation take place. Furthermore, we single out privacy concerns related to identity when used from trust and reputation systems.
We specifically focus on two vulnerabilities that current trust and reputation systems have: the change of identity and multiple identities problems. As a result, we provide a privacy-preserving solution to these vulnerabilities which integrates the explored relationships among identity, privacy, trust and reputation.

::

Francesca Toni, Imperial College London, United Kingdom

Title: Argumentation for trust in service-oriented architectures (20 minutes)

Abstract:

Argumentation can be used to model the uncertainty of decision-makers about the trustworthiness of service providers in service-oriented architectures. In particular, the existence (or not) of contractual agreements may support arguments in favour (or against) the trustworthiness of service providers or requestors. These arguments can be very useful in complementing information about past behaviour of these service providers or requestors. I will discuss a concrete method to combine arguments and past behaviour information and illustrate its advantages.

::

Denis Trcek, University of Ljubljana, Slovenia

Title: Qualitative judgment dynamics (15-20 minutes)

Abstract:

Trust management is turning out to be essential for advanced IT solutions. It was first addressed some ten years ago when the suggested approaches at that period were actually tackling security and not trust directly. Later, more advanced methodologies emerged that were based on Bayesian statistics. These were followed by Dempster-Shafer theory of evidence and its derivative, subjective algebra. In addition, some attempts were made that were based on game theory. However, trust is a manifestation of reasoning and judgment processes. It has to be treated in line with this fact and has to be adequately supported in virtual environments. Therefore, on the basis of experiments, a complimentary methodology called qualitative algebra has been developed, which addresses the core of trust phenomenon. It complements existing methodologies and enables a trust management in pervasive computing environments. It is anticipated that its development can be beneficial also other domains (in e.g. trading or negotiating scenarios), especially because trust is a complex issues and our approach enables its better understanding through simulations.

::

Joana Urbano, University of Porto, Portugal

Title: Using Contextual Information to Tune the Estimation of Trust (15 minutes)

Abstract:

Trust estimation is a fundamental step in several decision making processes that occur in electronic business environments, such as the selection of partners and the adaptive configuration of contracts. A relatively unexplored area in the trust research field is the usage of contextual information in order to approximate the trustworthiness estimation of a target agent to the specificities of the current business needs. In this talk, we start by presenting an overview on research work on context-aware trust systems, stressing out their positive characteristics and signaling some issues not addressed by them. Then, we present our own approach to trustworthiness estimation using contextual information. In particular, we describe the Contextual Fitness component, a module that uses machine learning techniques to measure the adequacy of the profile of the agents being evaluated to the specificities of the current business needs, therefore contextualizing the trustworthiness estimation for these agents. To conclude, we describe the experiments done to evaluate our approach and present our interpretation of the obtained results. We put forward these ideas in order to launch discussion on this topic.

::

ATTENDEES:

  • Eugenio Oliveira, University of Oporto.

::

SetPageWidth