Incremental Region Merging for Sparse Multidimensional Data Cubes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Multidimensional databases face inefficiencies due to sparsity, requiring costly re-computation of clustering for incremental data changes and relying on geometric distance, which is not user-centric or efficient.
Innovation Solution
Incremental region formation in multidimensional databases where new data points are merged with existing regions based on differences along a single dimension, allowing for user-interest-driven optimization without requiring a priori knowledge of the entire dataset.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If traditional clustering algorithms are used to form regions, then data density is improved, but computational cost increases significantly when data changes
Solution Approach 1:
The patent implements dynamic region formation where regions are created and merged incrementally as data points are added, rather than requiring complete re-clustering. The system maintains a forest of regions and dynamically merges them when new data points arrive, allowing the structure to adapt to changing data without recomputing from scratch.
Solution Approach 2:
The patent segments the data into multiple regions (forest of regions) that can be independently managed and merged. Each region contains a subset of data points, and the system merges regions based on similarity metrics rather than treating the entire dataset as a single unit, enabling incremental updates without global recomputation.
2Device complexity
If geometric distance is used for clustering, then region formation is simplified, but user interest in specific dimensions is not considered
Solution Approach 1:
The patent applies local quality by allowing different dimensions to have different weights in the similarity calculation. Instead of treating all dimensions equally as in traditional geometric distance, the system can emphasize dimensions that are more important to user interests, enabling tailored region formation that reflects specific analytical needs.
Solution Approach 2:
The patent changes the parameters used in distance calculation from standard geometric distance to a customized metric that incorporates user-interest-driven weighting. This allows the similarity measurement to adapt to different analytical scenarios by adjusting which dimensions are most significant for region merging decisions.
3Measurement precision
If complete data re-clustering is performed on updates, then region accuracy is maintained, but time consumption increases
Solution Approach 1:
The patent performs preliminary actions by pre-establishing a framework of regions and pre-defining merging criteria based on user interests. When updates occur, the system only needs to evaluate whether new data points should merge with existing regions using the pre-established criteria, rather than performing complete re-clustering, thus maintaining accuracy while reducing update time.
Data Source
AI summary
A solution is provided in which regions are formed incrementally as points are added to a database. Each time a point is added to the database, it may be assigned to a new region. This new region may then be compared against each of the current set of regions until one is found with which it can be successfully merged. Then, this newly merged region may be compared against each of the other regions until one is found with which it can be successfully merged. This process repeats until there are no more regions that can be merged. Determining whether or not a region can be merged with another region may involve determining whether or not the points in one region differ from the points in the other region only along one dimension. If so, then the point can be successfully merged with the region.