# US 2020/0265264 A1 · Hierarchical rule clustering

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

- Number: US 2020/0265264 A1
- Document ID: US20200265264A1
- Status: Published application
- Kind: A1
- Application number: 16/085,187
- Filed: 2016-04-29
- Published: 2020-08-20
- Earliest claimed: 2016-04-29
- Assignees: Hewlett Packard Enterprise Development LP
- Applicants: Hewlett Packard Enterprise Development LP
- Inventors: David Subiros Perez; Luis Miguel Vaquero Gonzalez
- Web page: https://patentgrep.com/patent/US20200265264A1

## CPC

- G06N 5/025: Extracting rules from data
- G06F 18/22: Matching criteria, e.g. proximity measures
- G06F 18/231: Hierarchical techniques, i.e. dividing or merging pattern sets so as to obtain a dendrogram
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)

## Abstract

Example embodiments relate to hierarchical rule clustering. The examples disclosed herein access information about a set of rules, where information for an individual rule comprises information about a set of hypershapes associated with the individual rule. A respective hypervolume for each set of hypershapes associated with each individual rule may be calculated based on the accessed information. A first rule and a second rule may be combined as a new individual rule in the set of rules based on overlaps between the calculated hypervolumes.

## Claims

### Claim 1

A system for hierarchical rule clustering, comprising a physical processor that implements machine readable instructions that cause the system to:
access information about a set of rules, where information for an individual rule comprises information about a set of hypershapes associated with the individual rule;
calculate, based on the accessed information, a respective hypervolume for each set of hypershapes associated with each individual rule; and
combine, based on overlaps between the calculated hypervolumes, a first rule and a second rule as a new individual rule in the set of rules.

### Claim 2 (depends on 1)

The system of claim 1; wherein the physical processor implements machine readable instructions that cause the system to:
determine, for each pair of rules in the set of rules, a respective pair overlap between calculated hypervolumes for each individual rule; and
responsive to determining that a first pair overlap between a first pair of rules is the highest determined overlap, combine the first rule and the second rule by:
combining respective sets of hypershapes associated with the first rule and the second rule in the first pair of rules.

### Claim 3 (depends on 2)

The system of claim 2, wherein the physical processor implements machine readable instructions that cause the system to:
determine, for each pair of rules, the respective pair overlap between hypervolumes by:
determining a respective Jaccard similarity for the respective sets of hypershapes associated with each pair of rules.

### Claim 4 (depends on 3)

The system of claim 3, wherein the physical processor implements machine readable instructions that cause the system to:
responsive to the Jaccard similarity being below a threshold, not combine the respective set of hypershapes associated with each individual rule.

### Claim 5 (depends on 1)

The system of claim 1, wherein the physical processor implements machine readable instructions that cause the system to:
determine, for each pair of rules, the respective pair overlap between hypervolumes by:
determining, in a separate thread for each pair of rules, a Jaccard similarity for the respective sets of hypershapes associated with each pair of rule.

### Claim 6 (depends on 2)

The system of claim 2, wherein the physical processor implements machine readable instructions that cause the system to:
normalize the calculated hypervolumes; and
determine, for each pair of rules in the set of rules, the respective pair overlap between normalized hypervolumes for each individual rule by:
determining, for each pair of rules in the set of rules, the respective pair overlap between normalized hypervolumes for each individual rule.

### Claim 7 (depends on 1)

The system of claim 1, wherein the physical processor implements machine readable instructions that cause the system to:
responsive to combining the first rule and the second rule as the new individual rule in the set of rules, calculate new respective hypervolumes for each set of hypershapes associated with each individual rule in the set of rules; and
combine a third rule and a fourth rule based on new overlaps between the calculated new hypervolumes.

### Claim 8

A method of hierarchical rule clustering, the method being implemented by a computer system comprising a physical processor implementing machine readable instructions, the method comprising:
iteratively combining pairs of rules in a set of rules by:
accessing information about the set of rules, where information for an individual rule comprises information about a set of hypershapes associated with the individual rule;
calculating, based on the accessed information, a respective hypervolume for each set of hypershapes associated with each individual rule; and
combining a first rule and a second rule as a new individual rule in the set of rules based on overlaps between the calculated hypervolumes.

### Claim 9 (depends on 8)

The method of claim 8, further comprising:
determining, for each pair of rules in the set of rules, a respective pair overlap between calculated hypervolumes for each individual rule; and
responsive to determining that a first pair overlap between a first pair of rules is the highest determined overlap, combining the first rule and the second rule by:
combining respective sets of hypershapes associated with the first rule and the second rule in the first pair of rules.

### Claim 10 (depends on 9)

The method of claim 9, further comprising:
determining, for each pair of rules, the respective pair overlap between hypervolumes by:
determining a respective Jaccard similarity for the respective sets of hypershapes associated with each pair of rules.

### Claim 11 (depends on 10)

The method of claim 10, further comprising:
determining, for each pair of rules, the respective pair overlap between hypervolumes by:
determining, in a separate thread for each pair of rules, a Jaccard similarity for the respective sets of hypershapes associated with each pair of rule.

### Claim 12 (depends on 10)

The method of claim 10, wherein iteratively combining pairs of rules in a set of rules further comprises;
sorting the respective Jaccard similarities for each pair of rules; and
iteratively combining pairs of rules in order of their corresponding respective Jaccard similarities.

### Claim 13 (depends on 8)

The method of claim 8, further comprising:
marking as non-modifiable hypershapes associated with the first rule, the second rule, and a subset of rules that neighbor the first rule and the second.

### Claim 14

A non-transitory machine-readable storage medium for hierarchical rule clustering, encoded with machine-readable instructions executable by the physical processor that cause the system to:
access information about a set of rules, where information for an individual rule comprises information about a set of hypershapes associated with the individual rule;
calculate, based on the accessed information, a respective hypervolume for each set of hypershapes associated with each individual rule;
determine, based on the calculated respective hypervolumes, a respective pair overlap for each pair of rules in the set of rules; and
combine, based on determined pair overlaps, a first rule and a second rule as a new individual rule in the set of rules.

### Claim 15 (depends on 14)

The non-transitory storage medium of claim 14, further comprising machine readable instructions that cause the system to:
determine, for each pair of rules, the respective pair overlap between hypervolumes by: determining a respective Jaccard similarity for the respective sets of hypershapes associated with each pair of rules; and
responsive to determining that a first pair overlap between a first pair of rules is the highest determined overlap, combine the first rule and the second rule by: combining respective sets of hypershapes associated with the first rule and the second rule in the first pair of rules.
