System for filtering spam messages based on user reputation

摘要

A system for updating filtering rules for messages (100) received from a plurality of users (160), the system comprising a processor, memory connected to the processor, and computer code loaded into the memory to to implement the following steps: storing filter rules for the plurality of users (160) in a filter rule database (140); distributing the filter rules to the plurality of users (160); computing (420) a reputation weight for each of the plurality of Users (160); wherein calculating the reputation weights of the users (160) comprises: setting rules defining user actions based on which the reputation weights are changed when a user takes the action, the defined user actions including reports that a message belongs to a certain category, and at least one action other than a report that a message belongs to a certain Category Belongs to; maintaining a reputation rules database (630) of user actions, comprising at least one of the following: increasing the level of heuristic analysis for checking files, increasing the update period, instructing the anti-spam filter and allowing a suspicious file to run, and modifying the user reputation weights accordingly; distributing rules based on data in the reputation rule database defining user actions necessary to register the user actions on the client side; registering user actions on the client side based thereon rules defining user actions; sending notifications of registered user actions; and calculating reputation weights based on the received notifications of user actions and changing the reputation weight of the corresponding user (160) based on the modifications according to the reputation rules in the reputation rule database (630); storing a reputation weight for each of the plurality of users (160 ) in a user reputation database (320); andreceiving (410) and processing user reports indicating that a message belongs to a particular category, the processing comprising: (i) calculating (430) a weight of the message in its category based on a plurality of reports selected from a plurality from users (160), and a reputation weight of these users (160), (ii) deciding (440) whether the message (100) belongs to the particular category, if the weight of the message exceeds a predefined threshold value; (iii ) Updating (445) the filter rules in the filter rule database based on the decision, and (iv) distributing (455) the updated filter rules from the filter rule database to the plurality of users.