Sanitization models and their limitations

This work explores issues of computational disclosure control. We examine assumptions in the foundations of traditional problem statements and abstract models. We offer a comprehensive framework, based on the notion of an inference game, that unifies various inference problems by parameterizing their problem spaces. This work raises questions regarding the significance of intractability results. We analyze common structural aspects of inference problems via case studies; these emphasize why explicit policies are needed to specify all social context and ethical values relevant to a problem instance.