Abstract: In the Map based publish/subscribe systems, most of the typical event matching approaches start from the published events, and then move through to looking for the matched subscriptions. Since there are always some redundant attributes in different events, the same attribute would probably exist in more than one event. When the number of published events is big, the same attributes in different events would be matched more than once with the constraints in the subscriptions, i.e., there are redundant matches during the event matching process. To alleviate this redundant matching problem, a new event matching algorithm which is based on the judgment of redundant attributes in different events is presented in this paper. By judging the redundancy of attributes, merging the event sets to eliminate the matching redundancy, and maintaining the constraints in subscriptions set into a multi-level index structure, this new event matching algorithm improves the matching efficiency in time complexity and also the maintainability in space complexity. Based on the above approach, a series of experiments with comparison are made. The experiment results clearly show that this event matching algorithm has higher efficiency compared with similar approaches when the number of events and subscriptions is big approaching hundreds or thousands.