One of the key challenges in operational trust management is to continually monitor the behavior of a node and update its trust score accordingly - evidently, both speed and accuracy is of great importance here. To achieve these goals, several papers have explored the concept of mutual revocation (sometimes termed suicide) wherein the trust value of both the accuser and the accused node are temporarily set to zero without involving a quorum. In this paper we explore a partial mutual revocation approach wherein we design a class of trust update functions to temporarily punish both the accuser and accused node (without involving a quorum) - however, the trust update function does not essentially set their trust values to zero; instead it partially lowers the trust values of both the accuser and the accused. In addition, we allow a trusted authority or a quorum may (periodically) review such partial mutual revocations and update the trust values of the accuser and the accused nodes accordingly (e.g., reward the accuser and punish the accused if the accusation was deemed true). We present a detailed design of the trust update functions for partial mutual revocation. Through both analysis and simulations, we evaluate the effectiveness of partial revocation under different attack strategies and report its performance in terms of revocation immediacy, revocation accuracy and abuse resistance.