How users’ participation affects reputation management systems: The case of P2P networks

Abstract

In this work human factor is explored by means of agent based simulation and analyzed in the framework of a reputation management system (RMS), within a peer-to-peer (P2P) network. Reputation is about evaluating an agent’s actions and other agents’ opinions about those actions, reporting on those actions and opinions, and reacting to that report, thus creating a feedback loop. This social mechanism has been successfully used to classify agents within normative systems. The systems rely on the feedbacks given by the members of the social network in which the RMS operates. Reputation can thus be seen as an endogenous and self produced indicator, created by the users for the users’ benefit. This implies that users’ participation and collaboration is a key factor for the effectiveness a RMS.

Introduction

The purpose of this work is to show, by means of simulation, that in such systems based on social networks and, in general, social links an recommendations, individuals’ effective participation in the process of assigning a “reputation value” to others is a key factor for the well being of the system itself. This is particularly true when the system is a wide one, in which few agents actually know each other and hence have to rely on others’ opinions about the unknown subjects whom they deal with.

In everyday’s life, when a choice subject to limited resources (like for instance money, time, and so on) must be done, due to the overwhelming number of possibilities that people have to choose from, something is needed to help them in making choices. People often follow the advice of others when it comes to which products to buy, which movies to watch, which music to listen, which websites to visit, and so on. This is a social attitude that uses others’ experience. The people usually base their decision about whether or not following the received advices partially upon the other person’s reputation in helping to find reliable and useful information, even with all the noise.

Using and building upon early collaboration filtering techniques, reputation management software gather ratings for people, companies, and information sources. Since this is a distributed way of computing reputation, it is implicitly founded on two main assumptions:

• The correctness of shared information.

• The participation of users to the system.

While the negation of the first could be considered as an attack to the system itself, performed by users trying to crash it, and its occurrence is quite rare, the second factor is often underestimated, when designing a collaborative RMS. Users without a vision of the macro level often use the system, but simply forget to collaborate, since this seems to cause a waste of time.

In this work a qualitative and, when possible, quantitative evaluation of the collaborative factor in RMSs is given, by means of an empirical analysis conducted via an agent based simulation. Thus, the main research question is: what’s the effectiveness of a RMS, when changing the collaboration rate coming from the involved users?

In order to answer this question, in the paper an agent based model is introduced, representing a peer-to-peer (P2P) network for file sharing. A basic RMS is applied to the system, in order to help users to choose the best peers to download from. In fact, some of the peers are malicious, and they try to exploit the way in which the P2P system rewards users for sharing files, by uploading inauthentic resources when they do not own the real ones. The model is described in detail and the results are evaluated through a multi-run ceteris paribus technique, in which only one setting is changed at a time. In particular, the most important parameters which will be compared, to evaluate the effectiveness of the RMS are: verification of the files, performed by the users and negative payoff, given in case a resource is reported as being inauthentic. The verification of the files, i.e. users’ the collaboration, is an exogenous and not controllable factor for the RMS, while the negative payoff is an endogenous and thus directly controllable variable, from the point of view of a RMS’s designer.

The P2P framework has been chosen since there are many works focusing on the reputation as a system to overcome the issue of inauthentic files, but, when evaluating the effectiveness of the system, the authors [8] usually refer to idealized situations, in which users always verify the files for authenticity, as soon as they start a download. This is obviously not the case in the real world: first of all, most resources require to be at least partially owned, in order to be checked. Besides, some users could simply decide not to check them for long time. Even worse, other users could simply forget about a downloaded resource and never check it. Last but not least, other users might verify it, but simply not report anything, if it’s not authentic.

Even if this work uses P2P networks as a scenario, the authors’ goal is not that of proposing a new RMS, or improving an existing algorithm for reputation management.

Besides, this paper deals with “normal” situations, i.e.: those happening in the majority of cases. Thus, deliberate attacks to a RMS would not be considered, and the users are supposed to give a right opinion, if and when they decide to share it with others. That “if and when”, and how much it affects the effectiveness of a RMS is indeed the main topic of this research.