Pre-computed impression lists for web advertising query resolution

The patent segments the large set of webpage attributes into multiple smaller subsets, where each subset contains a specific group of attributes. Instead of storing and searching all attributes simultaneously, the system divides them into manageable segments that can be processed independently, reducing search complexity while maintaining data completeness.

The patent performs preliminary actions by pre-computing and storing intermediate results for each attribute subset before the actual advertising query is executed. This allows the system to have attribute data ready in advance, organized in a way that facilitates efficient querying without requiring complex real-time processing of all attributes.

The patent segments complex queries with multiple criteria into several simpler sub-queries, each handling a subset of attributes. This allows the system to process each sub-query independently and efficiently, then combine the results to achieve the same accuracy as processing the complex query directly, but in less time.

The patent applies partial action by processing queries in stages rather than all at once. It first processes subsets of attributes separately to identify candidate web pages, then applies additional criteria in subsequent stages. This stepwise approach maintains identification accuracy while significantly reducing overall processing time compared to evaluating all criteria simultaneously.