In peer scoring, it is not possible to globally identify and remove network spammers. An alternative solution, which is explained in the following post, can do so and might interest the libp2p community: Privacy-preserving p2p economic spam protection