Speculative Filter Index: Automatic Bloom Filter and Zone Map Creation Based on Query Pattern Frequency Analysis

We describe the Speculative Filter Index (SFI) system, which automatically creates lightweight probabilistic filter structures (bloom filters and zone maps) based on observed query patterns without explicit user configuration. The system monitors query predicate patterns, tracks their frequency and selectivity, and automatically creates bloom filters for high-frequency equality predicates and zone maps for range predicates. Unused filters are automatically dropped after a configurable inactivity period.While automatic index recommendation is well-established (Oracle Auto Indexing 19c, Azure SQL Automatic Tuning, Microsoft AutoAdmin 1996, Database Cracking), the specific combination of targeting lightweight probabilistic structures (bloom filters + zone maps) rather than full B-tree indexes, selected by predicate type (equality vs. range), with automatic lifecycle management (creation and eviction), is a specific engineering synthesis. This disclosure establishes prior art for this approach, noting thatindividual components have extensive prior art.