On merging two trust-networks in one with bipolar preferences

In this paper, we study weighted trust-networks (but also unweighted), where each edge is associated with either a positive or a negative score. Hence, we consider a distrust relationship as well, allowing a user to rate poor experiences with other individuals in his web of acquaintances. We propose an algorithm to compose two of such networks in a single one, in order to merge the knowledge obtained in two different communities of individuals (possibly partially-overlapping), through two different trust management-systems. Our algorithm is based on semiring algebraic-structures, in order to have a parametric computational-framework. Such composition can be adopted whenever two trust-based communities (with the same scope) need to be amalgamated: for instance, two competitor-companies that need to unify the trust-based knowledge on their (sub-) suppliers.