# US 2003/0212692 A1 · In-database clustering

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2003/0212692 A1
- Document ID: US20030212692A1
- Status: Published application
- Kind: A1
- Application number: 10/424,761
- Filed: 2003-04-29
- Published: 2003-11-13
- Earliest claimed: 2002-05-10
- Assignees: Oracle International Corporation
- Inventors: Marcos M. Campos; Boriana L. Milenova; Mark A. McCracken
- Web page: https://patentgrep.com/patent/US20030212692A1

## CPC

- G06F 16/285: {Clustering or classification}
- G06F 18/2321: using statistics or function optimisation, e.g. modelling of probability density functions
- G06F 18/23213: with fixed number of clusters, e.g. K-means clustering
- Y10S 707/99942: Manipulating data structure, e.g. compression, compaction, compilation
- Y10S 707/99943: Generating database or data structure, e.g. via user interface
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- Y10S: TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS

## Abstract

A system, method, and computer program product for in-database clustering provides the capability to perform cluster analysis and provides improved performance in model building and data mining, good integration with the various databases throughout the enterprise, and flexible specification and adjustment of the models being built, but which provides data mining functionality that is accessible to users having limited data mining expertise and which provides reductions in development times and costs for data mining projects. A database management system for in-database clustering, comprises a first data table and a second data table, each data table including a plurality of rows of data, means for building a clustering model using the first data table, and means for applying the clustering model using the second data table to generate apply output data.

## Claims

### Claim 1

A database management system for in-database clustering, comprising:
a first data table and a second data table, each data table including a plurality of rows of data;
means for building a clustering model using the first data table; and
means for applying the clustering model using the second data table to generate apply output data.

### Claim 2 (depends on 1)

The database management system of claim 1, wherein:
the first data table and the second data table are the same data table.

### Claim 3 (depends on 1)

The database management system of claim 1, wherein:
the first data table and the second data table are different data tables.

### Claim 4 (depends on 1)

The database management system of claim 1, wherein the means for building a clustering model comprises:
a plurality of clustering model building routines.

### Claim 5 (depends on 4)

The database management system of claim 4, wherein the plurality of clustering model building routines comprises probabilistic model routine and rule generation routine require at least one clustering algorithm to work with:
a K-means model building routine, an Orthogonal Partitioning Clustering model building routine, a mixture model building routine, a probabilistic model building routine, and a rule generation routine.

### Claim 6 (depends on 5)

The database management system of claim 5, wherein the K-means model building routine comprises:
means for assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster; and
means for updating a centroid of at least one cluster.

### Claim 7 (depends on 6)

The database management system of claim 6, wherein the means for assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster comprises:
means for computing a distance metric to determine a distance between an input row of data in the first data table and each centroid of a cluster.

### Claim 8 (depends on 7)

The database management system of claim 7, wherein the means for updating a centroid of at least one cluster comprises:
means for replacing a current centroid of a cluster by a mean of rows of data in the first data table points assigned to the cluster.

### Claim 9 (depends on 5)

The database management system of claim 5, wherein the mixture model building routine comprises:
means for assigning a row of data in the first data table to a component based on a probability that the row of data belong to a distribution of the component; and
means for updating parameters of the distribution of the component using the row of data assigned to the component.

### Claim 10 (depends on 5)

The database management system of claim 5, wherein the Orthogonal Partitioning Clustering model building routine comprises:
means for generating a hierarchical grid-based clustering model.

### Claim 11 (depends on 5)

The database management system of claim 5, wherein the rule generation routine comprises:
means for extracting, for a group of clusters, a set of rules from information in histograms of the clusters.

### Claim 12 (depends on 11)

The database management system of claim 11, wherein the clusters are generated by the Orthogonal Partitioning Clustering model building routine.

### Claim 13 (depends on 4)

The database management system of claim 4, wherein the plurality of clustering model building routines comprises:
a K-means model building routine, an Orthogonal Partitioning Clustering model building routine, a mixture model building routine, a probabilistic model building routine, and a rule generation routine.

### Claim 14 (depends on 13)

The database management system of claim 13, wherein the K-means model building routine comprises:
means for assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster; and
means for updating a centroid of at least one cluster.

### Claim 15 (depends on 14)

The database management system of claim 14, wherein the means for assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster comprises:
means for computing a distance metric to determine a distance between an input row of data in the first data table and each centroid of a cluster.

### Claim 16 (depends on 15)

The database management system of claim 15, wherein the means for updating a centroid of at least one cluster comprises:
means for replacing a current centroid of a cluster by a mean of rows of data in the first data table points assigned to the cluster.

### Claim 17 (depends on 16)

The database management system of claim 16, wherein the mixture model building routine comprises:
means for assigning a row of data in the first data table to a component based on a probability that the row of data belong to a distribution of the component; and
means for updating parameters of the distribution of the component using the row of data assigned to the component.

### Claim 18 (depends on 16)

The database management system of claim 16, wherein the orthogonal partitioning clustering model building routine comprises:
means for generating a hierarchical grid-based clustering model.

### Claim 19 (depends on 16)

The database management system of claim 16, wherein the rule generation routine comprises:
means for extracting, for a group of clusters, a set of rules from information in histograms of the clusters.

### Claim 20 (depends on 19)

The database management system of claim 19, wherein the clusters are generated by the orthogonal partitioning clustering model building routine.

### Claim 21 (depends on 1)

The database management system of claim 1, wherein the means for applying the clustering model comprises:
a Naïve Bayes apply routine.

### Claim 22

In a database management system, a method for in-database clustering, comprising:
providing a first data table and a second data table, each data table including a plurality of rows of data;
building a clustering model using the first data table; and
applying the clustering model using the second data table to generate apply output data.

### Claim 23 (depends on 22)

The method of claim 22, wherein:
the first data table and the second data table are the same data table.

### Claim 24 (depends on 22)

The method of claim 22, wherein:
the first data table and the second data table are different data tables.

### Claim 25 (depends on 22)

The method of claim 22, wherein the step of building a clustering model comprises:
a plurality of clustering model building routines.

### Claim 26 (depends on 25)

The method of claim 25, wherein the plurality of clustering model building routines comprises at least one of:
a K-means model building routine, an Orthogonal Partitioning Clustering model building routine, a mixture model building routine, a probabilistic model building routine, and a rule generation routine.

### Claim 27 (depends on 26)

The method of claim 26, wherein the K-means model building routine comprises:
assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster; and
updating a centroid of at least one cluster.

### Claim 28 (depends on 27)

The method of claim 27, wherein the step of assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster comprises:
computing a distance metric to determine a distance between an input row of data in the first data table and each centroid of a cluster.

### Claim 29 (depends on 28)

The method of claim 28, wherein the step of updating a centroid of at least one cluster comprises:
replacing a current centroid of a cluster by a mean of rows of data in the first data table points assigned to the cluster.

### Claim 30 (depends on 26)

The method of claim 26, wherein the mixture model building routine comprises:
assigning a row of data in the first data table to a component based on a probability that the row of data belong to a distribution of the component; and
updating parameters of the distribution of the component using the row of data assigned to the component.

### Claim 31 (depends on 26)

The method of claim 26, wherein the Orthogonal Partitioning Clustering model building routine comprises:
generating a hierarchical grid-based clustering model.

### Claim 32 (depends on 26)

The method of claim 26, wherein the rule generation routine comprises:
extracting, for a group of clusters, a set of rules from information in histograms of the clusters.

### Claim 33 (depends on 32)

The method of claim 32, wherein the clusters are generated by the Orthogonal Partitioning Clustering model building routine.

### Claim 34 (depends on 25)

The method of claim 25, wherein the plurality of clustering model building routines comprises:
a K-means model building routine, an Orthogonal Partitioning Clustering model building routine, a mixture model building routine, a probabilistic model building routine, and a rule generation routine.

### Claim 35 (depends on 34)

The method of claim 34, wherein the K-means model building routine comprises:
assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster; and
updating a centroid of at least one cluster.

### Claim 36 (depends on 35)

The method of claim 35, wherein the step of assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster comprises:
computing a distance metric to determine a distance between an input row of data in the first data table and each centroid of a cluster.

### Claim 37 (depends on 36)

The method of claim 36, wherein the step of updating a centroid of at least one cluster comprises:
replacing a current centroid of a cluster by a mean of rows of data in the first data table points assigned to the cluster.

### Claim 38 (depends on 34)

The method of claim 34, wherein the mixture model building routine comprises:
assigning a row of data in the first data table to a component based on a probability that the row of data belong to a distribution of the component; and
updating parameters of the distribution of the component using the row of data assigned to the component.

### Claim 39 (depends on 34)

The method of claim 34, wherein the Orthogonal Partitioning Clustering model building routine comprises:
generating a hierarchical grid-based clustering model.

### Claim 40 (depends on 34)

The method of claim 34, wherein the rule generation routine comprises:
extracting, for a group of clusters, a set of rules from information in histograms of the clusters.

### Claim 41 (depends on 40)

The method of claim 40, wherein the clusters are generated by the Orthogonal Partitioning Clustering model building routine.

### Claim 42 (depends on 22)

The method of claim 22, wherein the step of applying the clustering model comprises:
a Naïve Bayes apply routine.

### Claim 43

A database management system for performing in-database clustering, comprising:
a processor operable to execute computer program instructions; and
a memory operable to store computer program instructions executable by the processor, for performing the steps of:
providing a first data table and a second data table, each data table including a plurality of rows of data;
building a clustering model using the first data table; and
applying the clustering model using the second data table to generate apply output data.

### Claim 44 (depends on 43)

The system of claim 43, wherein:
the first data table and the second data table are the same data table.

### Claim 45 (depends on 43)

The system of claim 43, wherein:
the first data table and the second data table are different data tables.

### Claim 46 (depends on 43)

The system of claim 43, wherein the step of building a clustering model comprises:
a plurality of clustering model building routines.

### Claim 47 (depends on 46)

The system of claim 46, wherein the plurality of clustering model building routines comprises at least one of:
a K-means model building routine, an Orthogonal Partitioning Clustering model building routine, a mixture model building routine, a probabilistic model building routine, and a rule generation routine.

### Claim 48 (depends on 47)

The system of claim 47, wherein the K-means model building routine comprises:
assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster; and
updating a centroid of at least one cluster.

### Claim 49 (depends on 48)

The system of claim 48, wherein the step of assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster comprises:
computing a distance metric to determine a distance between an input row of data in the first data table and each centroid of a cluster.

### Claim 50 (depends on 49)

The system of claim 49, wherein the step of updating a centroid of at least one cluster comprises:
replacing a current centroid of a cluster by a mean of rows of data in the first data table points assigned to the cluster.

### Claim 51 (depends on 47)

The system of claim 47, wherein the mixture model building routine comprises:
assigning a row of data in the first data table to a component based on a probability that the row of data belong to a distribution of the component; and
updating parameters of the distribution of the component using the row of data assigned to the component.

### Claim 52 (depends on 47)

The system of claim 47, wherein the Orthogonal Partitioning Clustering model building routine comprises:
generating a hierarchical grid-based clustering model.

### Claim 53 (depends on 47)

The system of claim 47, wherein the rule generation routine comprises:
extracting, for a group of clusters, a set of rules from information in histograms of the clusters.

### Claim 54 (depends on 53)

The system of claim 53, wherein the clusters are generated by the Orthogonal Partitioning Clustering model building routine.

### Claim 55 (depends on 46)

The system of claim 46, wherein the plurality of clustering model building routines comprises:
a K-means model building routine, an Orthogonal Partitioning Clustering model building routine, a mixture model building routine, a probabilistic model building routine, and a rule generation routine.

### Claim 56 (depends on 55)

The system of claim 55, wherein the K-means model building routine comprises:
assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster; and
updating a centroid of at least one cluster.

### Claim 57 (depends on 56)

The system of claim 56, wherein the step of assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster comprises:
computing a distance metric to determine a distance between an input row of data in the first data table and each centroid of a cluster.

### Claim 58 (depends on 57)

The system of claim 57, wherein the step of updating a centroid of at least one cluster comprises:
replacing a current centroid of a cluster by a mean of rows of data in the first data table points assigned to the cluster.

### Claim 59 (depends on 55)

The system of claim 55, wherein the mixture model building routine comprises:
assigning a row of data in the first data table to a component based on a probability that the row of data belong to a distribution of the component; and
updating parameters of the distribution of the component using the row of data assigned to the component.

### Claim 60 (depends on 55)

The system of claim 55, wherein the Orthogonal Partitioning Clustering model building routine comprises:
generating a hierarchical grid-based clustering model.

### Claim 61 (depends on 55)

The system of claim 55, wherein the rule generation routine comprises:
extracting, for a group of clusters, a set of rules from information in histograms of the clusters.

### Claim 62 (depends on 61)

The system of claim 61, wherein the clusters are generated by the Orthogonal Partitioning Clustering model building routine.

### Claim 63 (depends on 43)

The system of claim 43, wherein the step of applying the clustering model comprises:
a Naïve Bayes apply routine.

### Claim 64

A computer program product for performing in-database clustering in a database management system, comprising:
a computer readable medium;
computer program instructions, recorded on the computer readable medium, executable by a processor, for performing the steps of:
providing a first data table and a second data table, each data table including a plurality of rows of data;
building a clustering model using the first data table; and
applying the clustering model using the second data table to generate apply output data.

### Claim 65 (depends on 64)

The computer program product of claim 64, wherein:
the first data table and the second data table are the same data table.

### Claim 66 (depends on 64)

The computer program product of claim 64, wherein:
the first data table and the second data table are different data tables.

### Claim 67 (depends on 64)

The computer program product of claim 64, wherein the step of building a clustering model comprises:
a plurality of clustering model building routines.

### Claim 68 (depends on 67)

The computer program product of claim 67, wherein the plurality of clustering model building routines comprises at least one of:
a K-means model building routine, an Orthogonal Partitioning Clustering model building routine, a mixture model building routine, a probabilistic model building routine, and a rule generation routine.

### Claim 69 (depends on 68)

The computer program product of claim 68, wherein the K-means model building routine comprises:
assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster; and
updating a centroid of at least one cluster.

### Claim 70 (depends on 69)

The computer program product of claim 69, wherein the step of assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster comprises:
computing a distance metric to determine a distance between an input row of data in the first data table and each centroid of a cluster.

### Claim 71 (depends on 70)

The computer program product of claim 70, wherein the step of updating a centroid of at least one cluster comprises:
replacing a current centroid of a cluster by a mean of rows of data in the first data table points assigned to the cluster.

### Claim 72 (depends on 68)

The computer program product of claim 68, wherein the mixture model building routine comprises:
assigning a row of data in the first data table to a component based on a probability that the row of data belong to a distribution of the component; and
updating parameters of the distribution of the component using the row of data assigned to the component.

### Claim 73 (depends on 68)

The computer program product of claim 68, wherein the Orthogonal Partitioning Clustering model building routine comprises:
generating a hierarchical grid-based clustering model.

### Claim 74 (depends on 68)

The computer program product of claim 68, wherein the rule generation routine comprises:
extracting, for a group of clusters, a set of rules from information in histograms of the clusters.

### Claim 75 (depends on 74)

The computer program product of claim 74, wherein the clusters are generated by the Orthogonal Partitioning Clustering model building routine.

### Claim 76 (depends on 64)

The computer program product of claim 64, wherein the plurality of clustering model building routines comprises:
a K-means model building routine, an Orthogonal Partitioning Clustering model building routine, a mixture model building routine, a probabilistic model building routine, and a rule generation routine.

### Claim 77 (depends on 76)

The computer program product of claim 76, wherein the K-means model building routine comprises:
assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster; and
updating a centroid of at least one cluster.

### Claim 78 (depends on 77)

The computer program product of claim 77, wherein the step of assigning each of at least of portion of the plurality of rows of data in the first data table to a cluster comprises:
computing a distance metric to determine a distance between an input row of data in the first data table and each centroid of a cluster.

### Claim 79 (depends on 78)

The computer program product of claim 78, wherein the step of updating a centroid of at least one cluster comprises:
replacing a current centroid of a cluster by a mean of rows of data in the first data table points assigned to the cluster.

### Claim 80 (depends on 76)

The computer program product of claim 76, wherein the mixture model building routine comprises:
assigning a row of data in the first data table to a component based on a probability that the row of data belong to a distribution of the component; and
updating parameters of the distribution of the component using the row of data assigned to the component.

### Claim 81 (depends on 76)

The computer program product of claim 76, wherein the Orthogonal Partitioning Clustering model building routine comprises:
generating a hierarchical grid-based clustering model.

### Claim 82 (depends on 76)

The computer program product of claim 76, wherein the rule generation routine comprises:
extracting, for a group of clusters, a set of rules from information in histograms of the clusters.

### Claim 83 (depends on 82)

The computer program product of claim 82, wherein the clusters are generated by the Orthogonal Partitioning Clustering model building routine.

### Claim 84 (depends on 64)

The computer program product of claim 64, wherein the step of applying the clustering model comprises:
a Naïve Bayes apply routine.
