CONSTRUCTING MULTIDIMENSIONAL HISTOGRAMS FOR COMPLEX SPATIAL GEOMETRY OBJECTS

• US 20140052711A1
• Filed: 08/16/2012
• Published: 02/20/2014
• Est. Priority Date: 08/16/2012
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Anderson, Richard J., Hu, Ying, Ravada, Siva, Bamba, Bhuvan
• Granted Patent: US 8,812,488 B2
• Time in Patent Office: Days
• US Class Current: 707/718
• CPC Class Codes: G06F 16/29 Geographical information databases

Abstract

Techniques are described for generating histograms for a multidimensional space. In the presence of large spatial objects, fuzzy splitting techniques are utilized to recursively divide the multidimensional space into partitions, where a single spatial object may belong to multiple partitions. Large spatial objects are essentially broken down into smaller objects that may allow for more efficient partitioning of the multidimensional space. A count of spatial objects in each partition yields a spatial histogram. A spatial object that belongs to multiple partitions may have a weighted count for each of the multiple partitions, based on the extent to which the spatial object overlaps with each partition. Thus, an object that is split among a handful of partitions will only contribute a fraction of a count to each partition. Small partitions having relatively few objects are avoided by refusing to subdivide a partition whose members drop below a threshold number.

Claims

1. A method comprising:

identifying spatial objects in an multidimensional space, each of the spatial objects covering an extent of the multidimensional space;

dividing the multidimensional space into partitions;

for each particular partition of the partitions, estimating a weighted count of how many of the spatial objects overlap with the particular partition;

wherein estimating the weighted count for a first particular partition comprises weighting each particular spatial object in a first subset of the spatial objects based on how much each particular spatial object overlaps with the first particular partition, wherein the each particular spatial object in the first subset of the spatial objects only partially overlaps the first particular partition;

wherein the method is performed by one or more computing devices.

2. The method of claim 1, further comprising determining boundaries for the partitions based on the extents of the spatial objects, the determining comprising dividing each particular spatial object in the first subset of the spatial objects into two or more weighted subobjects that have midpoints in different partitions of the partitions.

3. The method of claim 2, further comprising:

identifying a target number of partitions;

calculating a threshold dimension size based at least in part on the size of the multidimensional space and the target number of partitions;

selecting which of the spatial objects to divide into weighted subobjects based at least in part on comparing the sizes of the spatial objects to the threshold dimension size.

4. The method of claim 2, further comprising:

for each object of the spatial objects, identifying a bounding geometry that encloses the object;

wherein determining the boundaries for the partitions based on the extents of the spatial objects comprises determining the boundaries based upon the bounding geometries.

5. The method of claim 1, further comprising:

for each object of the spatial objects, identifying a bounding geometry that encloses the object;

wherein estimating a count of how many of the spatial objects overlap with the particular partition comprises identifying how many of the objects have bounding geometries that overlap with the particular partition;

wherein weighting each particular spatial object in the first subset of the spatial objects based on how much each particular spatial object overlaps with the first particular partition comprises weighting each particular spatial object by what percentage of the bounding geometry for the particular spatial object overlaps with the particular partition.

6. The method of claim 1, wherein dividing the multidimensional space into the partitions comprises:

splitting an initial partition into a set of partitions; and

recursively splitting each particular partition in the set of partitions into additional partitions until a target number of partitions have been identified;

wherein splitting each particular partition comprises;

splitting the particular partition into two or more partitions whose boundaries overlap approximately equally-sized subsets of the spatial objects; and

identifying a particular set of spatial objects that overlap multiple partitions of the two or more partitions;

splitting each particular spatial object in the particular set of spatial objects into two or more weighted subobjects having midpoints in different partitions of the two or more partitions.

7. The method of claim 1, wherein dividing the multidimensional space into the partitions comprises:

splitting an initial partition into a set of partitions; and

recursively splitting each particular partition in the set of partitions into approximately equally-sized additional partitions until a target number of partitions have been identified;

wherein recursively splitting each particular partition comprises determining whether to recursively split a second particular partition of the equally-sized additional partitions based on how many of the spatial objects overlap the second particular partition.

8. The method of claim 1, further comprising:

storing a histogram describing the partitions and specifying the estimated count for each of the partitions;

using, by a query optimizer that is compiling a query, the histogram to calculate selectivity estimates for a table in which the spatial objects are stored;

wherein each of the spatial objects is specified by a different row of the table.

9. A method comprising:

dividing a multidimensional space into a set of one or more partitions;

selecting a particular partition from the set of one or more partitions, based on the particular partition having a largest size among partitions in the set of one or more partitions that overlap with more than a threshold number of spatial objects;

splitting the particular partition into multiple partitions and replacing the particular partition in the set of one or more partitions with the multiple partitions;

repeating at least the steps of selecting a particular partition and splitting the particular partition with respect to different partitions in the set of one or more partitions, until the set of one or more partitions includes a target number of partitions;

generating a histogram that indicates, for each particular partition of the set of one or more partitions, an estimated count of spatial objects that overlap with the particular partition;

using the estimated counts in the histogram to identify a query plan for executing a database query involving spatial objects in the multidimensional space;

wherein the method is performed by one or more computing devices.

10. The method of claim 9, further comprising:

upon splitting a first particular partition into a first set of multiple partitions, splitting a first set of spatial objects that overlap with two or more partitions in the first set of partitions into multiple spatial objects that are assigned weights based on size relative to each other, each of the multiple spatial objects residing entirely within a different partition of the multiple partitions;

calculating the estimated counts based in part on the assigned weights.

11. A method comprising:

dividing a multidimensional space into a set of one or more partitions;

selecting a particular partition from the set of one or more partitions, based on the particular partition having a largest number of spatial objects that overlap with the particular partition;

splitting the particular partition into multiple partitions and replacing the particular partition in the set of one or more partitions with the multiple partitions;

splitting each spatial object, in a first set of spatial objects that overlap with two or more partitions in the multiple partitions, into multiple spatial objects that are assigned weights based on size relative to each other, each of the multiple spatial objects residing entirely within a different partition of the multiple partitions;

repeating at least the steps of selecting a particular partition and splitting the particular partition with respect to different partitions in the set of one or more partitions, until the set of one or more partitions includes a target number of partitions;

generating a histogram that indicates, for each particular partition of the set of one or more partitions, an estimated count of spatial objects that overlap with the particular partition, wherein the estimated counts are based in part on the assigned weights;

using the estimated counts in the histogram to identify a query plan for executing a database query involving spatial objects in the multidimensional space;

wherein the method is performed by one or more computing devices.

12. One or more non-transitory media storing instructions which, when executed by one or more computing devices, cause performance of:

identifying spatial objects in an multidimensional space, each of the spatial objects covering an extent of the multidimensional space;

dividing the multidimensional space into partitions;

for each particular partition of the partitions, estimating a weighted count of how many of the spatial objects overlap with the particular partition;

wherein estimating the weighted count for a first particular partition comprises weighting each particular spatial object in a first subset of the spatial objects based on how much each particular spatial object overlaps with the first particular partition, wherein the each particular spatial object in the first subset of the spatial objects only partially overlaps the first particular partition.

13. The one or more non-transitory media of claim 12, wherein the instructions, when executed by the one or more computing devices, further cause performance of determining boundaries for the partitions based on the extents of the spatial objects, the determining comprising dividing each particular spatial object in the first subset of the spatial objects into two or more weighted subobjects that have midpoints in different partitions of the partitions.

14. The one or more non-transitory media of claim 13, wherein the instructions, when executed by the one or more computing devices, further cause performance of:

identifying a target number of partitions;

calculating a threshold dimension size based at least in part on the size of the multidimensional space and the target number of partitions;

selecting which of the spatial objects to divide into weighted subobjects based at least in part on comparing the sizes of the spatial objects to the threshold dimension size.

15. The one or more non-transitory media of claim 13, wherein the instructions, when executed by the one or more computing devices, further cause performance of:

for each object of the spatial objects, identifying a bounding geometry that encloses the object;

wherein determining the boundaries for the partitions based on the extents of the spatial objects comprises determining the boundaries based upon the bounding geometries.

16. The one or more non-transitory media of claim 12, wherein the instructions, when executed by the one or more computing devices, further cause performance of:

for each object of the spatial objects, identifying a bounding geometry that encloses the object;

wherein estimating a count of how many of the spatial objects overlap with the particular partition comprises identifying how many of the objects have bounding geometries that overlap with the particular partition;

wherein weighting each particular spatial object in the first subset of the spatial objects based on how much each particular spatial object overlaps with the first particular partition comprises weighting each particular spatial object by what percentage of the bounding geometry for the particular spatial object overlaps with the particular partition.

17. The one or more non-transitory media of claim 12, wherein dividing the multidimensional space into the partitions comprises:

splitting an initial partition into a set of partitions; and

recursively splitting each particular partition in the set of partitions into additional partitions until a target number of partitions have been identified;

wherein splitting each particular partition comprises;

splitting the particular partition into two or more partitions whose boundaries overlap approximately equally-sized subsets of the spatial objects; and

identifying a particular set of spatial objects that overlap multiple partitions of the two or more partitions;

splitting each particular spatial object in the particular set of spatial objects into two or more weighted subobjects having midpoints in different partitions of the two or more partitions.

18. The one or more non-transitory media of claim 12, wherein dividing the multidimensional space into the partitions comprises:

splitting an initial partition into a set of partitions; and

recursively splitting each particular partition in the set of partitions into approximately equally-sized additional partitions until a target number of partitions have been identified;

wherein recursively splitting each particular partition comprises determining whether to recursively split a second particular partition of the equally-sized additional partitions based on how many of the spatial objects overlap the second particular partition.

19. The one or more non-transitory media of claim 12, wherein the instructions, when executed by the one or more computing devices, further cause performance of:

storing a histogram describing the partitions and specifying the estimated count for each of the partitions;

using, by a query optimizer that is compiling a query, the histogram to calculate selectivity estimates for a table in which the spatial objects are stored;

wherein each of the spatial objects is specified by a different row of the table.