Can you trust the user?: Collaborative trust estimation model for recommendations

Abstract

The explosive popularity of social networking services such as Twitter and Facebook has made it common to communicate with unknown people via the Internet. Further, services based on a “sharing economy” such as Airbnb and Uber have gained popularity and increased opportunities to connect with strangers. Users would like to determine whether the opinions of other users can be trusted. However, it is sometimes difficult to do so, and the judgment of other users is not necessarily correct. We propose a model for predicting the trust of unknown users by using information of users known to a target user. We predicted the trust of reviewers on an online review site based on the proposed model and recommended items based on collaborative filtering (CF) using the obtained trust of the reviewers. The experimental results showed that the recommendation accuracy with the trust was higher than that with general CF.