Multi-dimensional Data Query via Spatial Subspace Clustering

The patent divides each data block into multiple subspace objects based on spatial clustering, where each subspace object represents a specific data range within the block. This segmentation allows the query system to evaluate smaller, more precise units rather than treating entire blocks as atomic units, thereby reducing the number of blocks that need to be read while maintaining simple selection logic.

The patent introduces a new dimension of granularity by creating subspace objects within blocks. Instead of operating only at the block level, the system now operates at the subspace level, adding a dimensional layer between blocks and individual data points. This enables more precise filtering without increasing the complexity of the selection process.

The patent performs spatial clustering and constructs subspace objects during the data loading phase, before any queries are executed. This preliminary action pre-computes the spatial distribution information and stores it in the database catalog, so that during query processing, the system only needs to retrieve and use this pre-computed information without performing complex calculations in real-time.

The patent introduces subspace objects as an intermediary layer between raw data blocks and query processing. These subspace objects serve as mediators that encapsulate spatial distribution information, allowing the query system to interact with blocks through this intermediate representation rather than directly handling raw block data, thereby simplifying the query process while improving efficiency.