Systems and methods for optimizing decision graph collaborative filtering

A data analysis system that facilitates determination of user preferences, comprising:

• means for compiling default scores for variables, exclusive of invalidated variables, of decision graphs/trees, the invalidated variables comprising a variable-invalidation list of each variable associated with a plurality of variables that it invalidates, the variable-invalidation list employed to construct an invalidation list by taking a union of a set of evidence and the variable-invalidation list;
• and
• means for generating a top-n prediction list to facilitate item prediction, the top-n prediction list is seeded by the default scores for the variables.

A data analysis system, comprising:

• means for compiling invalidation scores for variables of decision graphs/trees, the means for compiling employing an variable-invalidation list of each variable associated with variables that it invalidates and a set of evidence to create an invalidation list by taking a union of the variable-invalidation list and the set of evidence;
• and
• means for generating a top-n prediction list to facilitate item prediction, the top-n prediction list is seeded by non-default scores of the variables in the invalidation list, the top-n prediction list employed to produce recommendations of user preferences.