Database Query Processing with RID List Merging

The patent segments the dimension filtering process by creating separate RID lists for each dimension table instead of forming a complete Cartesian product. Each RID list contains only the relevant record identifiers for that dimension, allowing independent filtering without materializing all possible combinations. This segmentation reduces memory consumption while maintaining the ability to consolidate filtering results through intersection operations on the RID lists.

The patent extracts only the necessary record identifiers (RIDs) from each dimension table that match the query criteria, rather than processing entire tables or their Cartesian product. By extracting and storing RIDs in separate lists, the system consolidates dimension filtering through intersection operations on these compact lists, significantly reducing the computational burden and improving query processing performance while maintaining filtering versatility.

The patent introduces RID lists as intermediary structures that mediate between normalized dimension tables and the fact table. Instead of performing multiple sequential joins between normalized dimension tables, the system generates RID lists from each dimension table independently, then uses these RID lists to directly access relevant records in the fact table. This intermediary approach maintains the benefits of normalization (reduced redundancy) while simplifying the join operation complexity.