Generally described, a method and system for providing a flexible rule-based infrastructure for a self-policing discussion board is provided. More specifically, a method and system is provided for processing reports for messages posted in a discussion board and determining whether each report is accurate based on predetermined criteria. The predetermined criteria are defined by a set of report handling rules which evolve over time by learning from previously processed activity reports. The set of report handling rules also defines an appropriate action to take with regards to an accused message as response to the report when the predetermined criteria are satisfied.