Eleni Koutrouli

Reputation Management Survey

Electronic markets, distributed peer-to-peer applications and other forms of online collaboration are all based on mutual trust, which enables transacting peers to overcome the uncertainty and risk inherent in the environment. Reputation systems provide essential input for computational trust as predictions on future behaviour based on the past actions of a peer In order to analyze the maturity of current reputation systems, we compare eleven reputation systems within a taxonomy of the credibility aspects of a reputation system. The taxonomy covers three topics: 1) the creation and content of a recommendation, 2) the selection and use of recommenders, and 3) the interpretation and reasoning applied to the gathered information. Although we find it possible to form a trusted reputation management network over an open network environment, there are still many regulatory and technical obstacles to address. This survey reveals various good mechanisms and methods used, but the area still requires both a) formation of standard mechanisms and metrics for reputation system collaboration and b) standard metainformation of right granularity for evaluating the credibility of reputation information provided

Taxonomy of attacks and defense mechanisms in P2P reputation systems—Lessons for reputation system designers

A B S T R A C T Robust and credible reputation systems are essential for the functionality of Peer-to-Peer (P2P) applications. However, they themselves are susceptible to various types of attacks. Since most current efforts lack an exploration of a comprehensive adversary model, we try to fi ll in this gap by providing a thorough view of the various credibility threats against a decentralized reputation system and the respective defense mechanisms. Therefore, we explore and classify the types of potential attacks against reputation systems for P2P applications. We also study and classify the defense mechanisms which have been proposed for each type of attack and identify confl icts between defense mechanisms and/or desirable characteristics of credible reputations systems. We fi nally propose a roadmap for reputation system designers on how to use the results of our survey for the design of robust reputation systems for P2P applications. c

Reputation-Based trust systems for p2p applications: design issues and comparison framework

In Peer-to-Peer (P2P) computing area trust issues have gained focus as a result of the decentralized nature of P2P systems where autonomous peers interact with each other without relying on any central authority. There is, thus, the need of a trust system to ensure a level of robustness against malicious nodes. Various reputation-based trust models have been proposed for P2P systems which use similar concepts but focus on different aspects and address different set of design issues. As a result, there is a clear need to investigate the design aspects of reputation-based trust systems that could be deployed in P2P applications. In this paper we present the basic elements and design issues of such systems and compare representative approaches, aiming at supporting the design of reputation systems suitable for particular P2P applications.

Reputation Systems Evaluation Survey

Various reputation systems have been proposed for a broad range of distributed applications, such as peer-to-peer, ad-hoc, and multiagent systems. Their evaluation has been mostly based on proprietary methods due to the lack of widely acceptable evaluation measures and methodologies. Differentiating factors in various evaluation approaches include the evaluation metrics, the consideration of the dynamic behavior of peers, the use of social networks, or the study of resilience to specific threat scenarios. The lack of a generally accepted common evaluation framework hinders the objective evaluation and comparison of different reputation systems. Aiming at narrowing the gap in the research area of objective evaluation of reputation systems, in this article, we study the various approaches to evaluating and comparing reputation systems, present them in a taxonomy, and analyze their strengths and limitations, with special focus on works suggesting a Common Evaluation Framework (CEF). We identify the challenges for a widely accepted CEF that enables testing and benchmarking of reputation systems, and we present the required properties for such a CEF; we also present an analysis of current CEF-related works in the context of the identified properties and our related proposals.

Credibility Enhanced Reputation Mechanism for Distributed E-communities

P2P communities are online communities of entities which offer services to each other without a central administration. Community members that need a specific service have to choose the transaction partner which they believe that will provide them with a required service of the expected quality, thus, they need mechanisms to support trust decisions regarding who they will transact with. P2P reputation systems provide trust mechanisms for P2P communities. Due to their decentralized and social nature, they are vulnerable to various types of attacks which distort their credibility and effectiveness. In our paper we first identify the credibility factors of a P2P reputation system. Then we describe a dynamic reputation mechanism for P2P communities, which integrates some of these credibility factors and which can be used as part of a comprehensive reputation system enhancing its robustness against some types of attacks. Simulation results show the effectiveness of our mechanism in relation to other reputation mechanisms which do not integrate such factors.

Interoperability and eservices

eServices are the building blocks for loosely-coupled, distributed applications based on the Service Oriented Architecture (SOA) principles. One of the major benefits they offer is interoperability both between components of service oriented systems and between different systems. Still, the variety and diversity of implementations and interpretations of SOA and the vast amount of emerging standards hinder interoperability. This paper examines interoperability requirements and related issues in the three major eServices categories: Web, Grid and P2P services. Our aim is to provide the basis for a roadmap towards improving interoperability of eServices.

Credible recommendation exchange mechanism for P2P reputation systems

The value of Reputation Systems has been widely recognized for P2P and other distributed applications. This value depends on how credible reputation systems can be, as they face a lot of attacks and various types of misbehavior. Recommendation free riding and badmouthing are two of the most important problems in reputation systems. Incentives constitute an important tool for alleviating such problems. We propose a credit-based recommendation exchange mechanism which provides incentives for honest participation in the reputation system. Our mechanism uses payments for recommendations, which depend on the trustworthiness of peers regarding the recommendations they give. No currency is circulated in the network. Peers have accounts, which are credited or debited according to recommendation transfers, while payment values are defined according to the recommendation trustworthiness of peers. We analyze the proposed payment scheme and provide simulation results to validate our analysis. We also discuss ways of secure recommendation exchange and related challenges.

Reputation Mechanisms in on-line Social Networks: The case of an Influence Estimation System in Twitter

Online social networks have changed the way people communicate, create content and share it via online social actions. In order for users to engage in a social network application, they need mechanisms that will help them to find other users that they can trust in some context and also qualitative and useful content. Reputation systems offer such mechanisms and are thus essential in Social Networks (SNs). Due to their open and social nature and to their different objectives, reputation systems for SN-based applications comprise of various different mechanisms which fulfill the different objectives in the specific application context. There is thus the need for a systematic study of the specific aspects of these systems, which can provide insights for the design of reputation systems in specific application contexts. In our paper we present a taxonomy of reputation systems for various SN applications based on their specific aspects. We then use this taxonomy to design a reputation system for microblogging systems and present our implementation for a reputation system for Twitter which estimates a specific notion of reputation; the influence of users and of specific subjects of discussion.

P2P Reputation Systems Credibility Analysis: Tradeoffs and Design Decisions

Peer-to-Peer (P2P) systems are distributed self-organized systems without a controlling entity where peers need to make trust decisions regarding who they will transact with. P2P reputation systems exploit past transactional information to provide members of P2P networks with measures which will help them make such trust decisions. However, these systems are vulnerable themselves to various types of attacks which can distort their functionality and credibility. A number of defense mechanisms have been proposed to counteract specific attacks, but research lacks an exploration of how these mechanisms can be used together so as to provide the highest degree of credibility. In this paper, after presenting the basic attacks and defense mechanisms concerning P2P reputation systems, we analyze the tradeoffs between different credibility enhancing mechanisms and ways to achieve the right balances. Our aim is to help reputation systems designers to incorporate in their systems the right mechanisms or the right combinations of mechanisms against the threats of the specific reputation systems.

Finding Topic-Specific Trends and Influential Users in Social Networks

Social networks (SNs) have become an integral part of contemporary life, as they are increasingly used as a basic means for communication with friends, sharing of opinions and staying up to date with news and current events. The general increase in the usage and popularity of social media has led to an explosion of available data, which creates opportunities for various kinds of utilization, such as predicting, finding or even creating trends. We are thus interested in exploring the following questions: (a) Which are the most influential - popular internet publications posted in SNs, for a specific topic? (b) Which members of SNs are experts or influential regarding a specific topic? Our approach towards answering the above questions is based on the functionality of hashtags, which we use as topic indicators for posts, and on the assumption that a specific topic is represented by multiple hashtags. We present a neighborhood-based recommender system, which we have implemented using collaborative filtering algorithms in order to (a) identify hashtags, urls and users related with a specific topic, and (b) combine them with SN-based metrics in order to address the aforementioned questions in Twitter. The recommender system is built on top of Apache Spark framework in order to achieve optimal scaling and efficiency. For the verification of our system we have used data sets mined from Twitter and tested the extracted results for influential users and urls concerning specific topics in comparison with the influence scores produced by a state of the art influence estimation tool for SNs. Finally, we present and discuss the results regarding two distinct topics and also discuss the offered and potential utility of our system.