Title: A referral-functional trust perspective for enhanced recommendations

Authors: Deepa Anand; Kamal K. Bharadwaj

Addresses: Department of Computer Science, Christ University, Hosur Road, Bangalore, 560029, India ' School of Computer and System Sciences, Jawaharlal Nehru University, Delhi, 110067, India

Abstract: There has been considerable interest recently in trust-enhanced recommender systems (RS). Trust is employed as an aid to overcome the sparsity problem in recommender systems (RS) by utilising the web of trust to discover proximity between users who can otherwise not be deemed similar. Such similar user discovery process utilises trust transitivity to infer the quantum of trust between a pair of users who do not have a trust link between them but are connected through the network. This enables broadening the set of users contributing to the recommendation. However, most trust-based approaches for RS do not distinguish between the trust in a party to suggest items (functional trust) and the trust in a party to recommend good recommenders (referral trust). We propose a referral-functional trust approach to RS suggesting different ways of modelling referral trust for a binary trust network and incorporating it in the trust propagation scheme to obtain better quality recommendations. Our approaches utilise the quality of the neighbourhood offered through a user's connections to estimate the user's referring capacity. The experimental results, showing comparison with existing schemes for trust inference, have established the effectiveness of our approach.

Keywords: recommender systems; functional trust; computational intelligence; referral trust; collaborative filtering; recommendations; modelling; trust inference.

DOI: 10.1504/IJCSYSE.2015.067755

International Journal of Computational Systems Engineering, 2015 Vol.2 No.1, pp.1 - 10

Published online: 31 Mar 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article