Performance of incremental update in database rule processing

The PARADISER (parallel and distributed environment for rules) system has been designed to provide efficient rule program evaluation on large databases. In PARADISER, an incremental update facility is provided for efficiently maintaining the consistency of a database rule inference process while changes to the initial set of facts occurs front some external source. We present measured performance results of incremental update in PARADISER by studying the behavior of two rule-based benchmark programs. The results we present indicate that incremental update can provide substantial performance gains when judiciously applied in appropriate contexts.<<ETX>>