Triggers in active database systems execute predefined actions if a trigger condition is satisfied. Since a trigger condition is evaluated whenever a corresponding event occurs, the performance of active database systems is highly affected by condition evaluation methods. In this thesis, we propose a new effective condition evaluation method based on discrimination networks. Our proposed networks accommodate general database operations, and also aggregation functions and differential tables such as NEW and OLD that have not been much considered in the past. The proposed condition evaluation method avoids redundant execution and reduces search ranges.
