Predicting User Reputation through Centrality Metrics

What is it about?

User reputation is a critical success factor of many Web-based platforms like e-marketplaces, product review Web sites, and question-and-answering systems. The most reputable users are those ones who post the most helpful reviews according to the opinion of the members of their community. Unfortunately, users generally post very few reviews and/or ratings on items they bought or consumed (like movies or music) and this makes the identification of highly reputable users hard. To effectively spot reputable users, we propose to leverage the “who-trusts-whom” network—known as a trust network— available in many platforms on the Social Web and to apply suitable centrality metrics (like Pagerank and Beetwenness centrality) to detect the most reputable users. Experiments on some large datasets provide evidence that we can leverage trust relationships to spot those users producing the most helpful reviews for the whole community.

Featured Image

Why is it important?

Our paper provides an empirical evidence that there exists a correlation between the position a user occupies in the trust network and her/his ability of producing helpful reviews. Therefore, we can benefit from the network of trust relationships to find out helpful reviewers even if few ratings and/or reviews are available. Finally, we show that Eigenvector Centrality performs the best -- in terms of the ability of correctly predicting the most helpful reviewers -- among centrality metrics we surveyed.

Perspectives