You should deduplicate and offer a cluster view with similarity distance