Measuring quality, reputation and trust in online communities

In the Internet era the information overload and the challenge to detect quality content has raised the issue of how to rank both resources and users in online communities. In this paper we develop a general ranking method that can simultaneously evaluate users’ reputation and objects’ quality in an iterative procedure, and that exploits the trust relationships and social acquaintances of users as an additional source of information. We test our method on two real online communities, the EconoPhysics forum and the Last.fm music catalogue, and determine how different variants of the algorithm influence the resultant ranking. We show the benefits of considering trust relationships, and define the form of the algorithm better apt to common situations.

💡 Research Summary

The paper addresses a fundamental problem in modern online platforms: how to rank both the content (objects) and the contributors (users) when faced with massive, noisy data. The authors propose a unified iterative ranking algorithm that simultaneously computes a reputation score for each user (R) and a quality score for each object (Q). The core of the method is a pair of mutually reinforcing update equations reminiscent of the HITS algorithm, but extended to incorporate explicit trust relationships among users.

Formally, let (w_{ui}) denote the weight of user (u)’s interaction with object (i) (e.g., a rating, a comment, or a play count). The object quality at iteration (t+1) is updated as a weighted average of the reputations of the users who interacted with it: \