Multi-faceted Trust-based Collaborative Filtering

Abstract

Many collaborative recommender systems leverage social correlation theories to improve suggestion performance. However, they focus on explicit relations between users and they leave out other types of information that can contribute to determine users' global reputation; e.g., public recognition of reviewers' quality.We are interested in understanding if and when these additional types of feedback improve Top-N recommendation. For this purpose, we propose a multi-faceted trust model to integrate local trust, represented by social links, with various types of global trust evidence provided by social networks. We aim at identifying general classes of data in order to make our model applicable to different case studies. Then, we test the model by applying it to a variant of User-to-User Collaborative filtering (U2UCF) which supports the fusion of rating similarity, local trust derived from social relations, and multi-faceted reputation for rating prediction. We test our model on two datasets: the Yelp one publishes generic friend relations between users but provides different types of trust feedback, including user profile endorsements. The LibraryThing dataset offers fewer types of feedback but it provides more selective friend relations aimed at content sharing. The results of our experiments show that, on the Yelp dataset, our model outperforms both U2UCF and state-of-the-art trust-based recommenders that only use rating similarity and social relations. Differently, in the LibraryThing dataset, the combination of social relations and rating similarity achieves the best results. The lesson we learn is that multi-faceted trust can be a valuable type of information for recommendation. However, before using it in an application domain, an analysis of the type and amount of available trust evidence has to be done to assess its real impact on recommendation performance.

A compositional model of multi-faceted trust for personalized item recommendation

“…This work extends the preliminary multi-faceted trust model presented in (Mauro et al, 2019) as follows: firstly, we integrate the facets of trust in Ma-trix Factorization, instead of using a K-Nearest Neighbors model. Secondly, we perform more detailed and extensive experiments to evaluate recommendation performance: (i) we analyze the impact of the facets of trust on recommendation by tuning the components of MTM in a finer-grained way;…”

Personalized multi-faceted trust modeling to determine trust links in social media and its potential for misinformation management

“…Inspired by work in the social sciences which have outlined the numerous variables which influence the formation of trust relationships [30], MFTM incorporates arbitrarily many indicators of trustworthiness into a single (optionally context-dependent) trustworthiness score. Operationalizing this core idea for trust and social tie prediction has been proposed by multiple researchers (e.g., [9,13,22,26,29]). As is evident in these works, there is little agreement over whether this technique should be called multi-dimensional, multi-faceted or composite trust modeling, and this confusion has likely led to some difficulty in coordinating efforts in this research direction.…”

“…A rising trend in this field is to validate models by applying their predictions to a recommendation task (e.g., [9,29]): that is, using the trust model to predict novel trust links, ˆ , in a multiagent system (MAS), then feeding those predicted links into a trust-aware item recommendation system. These trust-aware recommender systems incorporate both user-item rating behavior and user-user social/trust connections to better recommend items by leveraging the fact that social/trust connections exert influence on the preferences of agents (e.g., you are more likely to watch/enjoy a film a trusted friend recommends).…”

“…One promising new direction has been to recognize that multiple features of the data may be relevant, and thus that a proper weighting of these different contributing factors, when reasoning about trustworthiness, is important. This is the basic premise of the very novel pursuit known as multi-faceted trust modeling [9,29].…”

Personalized Prediction of Trust Links in Social Networks (Student Abstract)

“…In step 2, 18 features were extracted from the profiles and histories of the Yelp users. These features combined generic and domain specific perspectives, taking inspiration from the works of (Mauro, Ardissono, and Hu 2019) and (Fang, Guo, and Zhang 2015) (and moving beyond these solutions by taking a data driven approach to MFTM and employing more than a small set of general purpose indicators). One of the principal advantages of MFTM is its capability to process many features.…”