A collection of 1 billion publicly available web services can form an internet-scale infrastructure for building diverse applications. For a given application, selection of services and service providers from this collection becomes important and reputation is recognized as a key factor for this purpose. However, current reputation systems are limited in their ability to exchange reputation information between heterogeneous systems. To facilitate meaningful exchange and reuse of reputation information and for the overall determination of reputation, we identify the need to infer and explicate rationale for ratings. We present our knowledge based approach to inferring and explicating rationale for ratings. We show that this approach facilitates detection of deception and collusion, user preferences elicitation, explication of rationale behind user ratings and generation of personalized service recommendations.