Combining Left and for Matching a Large Num

摘要: In systems which learn a large number of rules (productions), it is important to match the efficiently, in order avoid machine learning utility probZem. So we need algorithms that scale well with productions system. (Doorenbos 1993) introduced right unlinking as way improve scalability Rete algorithm. This paper introduces symmetric optimization, left unlinking, and demonstrates makes on an even larger class systems. Unfortunately, when are combined same system, they can interfere each other. We give particular combine them prove minimizes this interference, analyze worst-case remaining interference. Finally, present empirical results showing interference very small practice, combination allows five our seven testbed over 100,000 without incurring significant increase cost.’