ViewpointS enables representation and storage of individual viewpoints in a shared knowledge graph. Knowledge providers (i.e., agents) express their individual opinions by emitting viewpoints on the semantic similarity or proximity between resources of the knowledge graph which can either be agents, documents (i.e., knowledge supports) or concepts (i.e., descriptors). We first briefly recall the ViewpointS knowledge representation formalism and discuss the genericity it enables in terms of semantic distance computation. In this paper, we benchmark the ViewpointS approach against other classic semantic distances (graph based or information content based) on a WordNet experiment. Our goal is to demonstrate the value of keeping the subjectivity of the represented knowledge, while having a generic approach that can handle any kind of knowledge and compute similarity between any kinds of objects.