# US 2022/0036006 A1 · Feature vector generation for probabalistic matching

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

- Number: US 2022/0036006 A1
- Document ID: US20220036006A1
- Status: Published application
- Kind: A1
- Application number: 16/942,925
- Filed: 2020-07-30
- Published: 2022-02-03
- Earliest claimed: 2020-07-30
- Assignees: International Business Machines Corporation
- Applicants: International Business Machines Corporation
- Inventors: Soma Shekar Naganna; Abhishek Seth; Neeraj Ramkrishna Singh
- Web page: https://patentgrep.com/patent/US20220036006A1

## CPC

- G06F 40/295: Named entity recognition
- G06F 16/285: {Clustering or classification}
- G06F 16/3347: {using vector based model}
- G06F 16/35: Clustering; Classification
- G06F 18/211: Selection of the most significant subset of features
- G06F 21/6245: {Protecting personal data, e.g. for financial or medical purposes}
- G06F 40/30: Semantic analysis
- G06F 40/279: Recognition of textual entities
- G06N 20/00: Machine learning
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

A computer-implemented method increases the efficiency of matching records from two sources. The method includes identifying a first source and a second source wherein each of the sources include one or more records and each record includes one or more attributes. The method further includes determining, based on a corpus, the one or more attributes and generating, based on the attributes, a set of feature vectors which vectors represent the one or more attributes. The method includes comparing each record in the first source against each record in the second source. The method further includes generating, in response to the comparing, a link confidence. The method also includes linking, in response to the link confidence being above a linking threshold, the associated records. The method includes determining a first feature vector of the set of feature vectors used in the linking, and outputting a set of results.

## Claims

### Claim 1

A computer-implemented method comprising:
identifying a first source and a second source wherein the first source and the second source include one or more records and each record includes one or more attributes;
determining, based on a corpus, the one or more attributes;
generating, based on the attributes, a set of feature vectors, wherein the set of feature vectors represent the one or more attributes;
comparing each record in the first source against each record in the second source;
generating, in response to the comparing, a link confidence;
linking, in response to the link confidence being above a linking threshold, a first record from the first source and a second record from the second record;
determining a first feature vector of the set of feature vectors used in the linking; and
outputting a set of results.

### Claim 2 (depends on 1)

The method of claim 1, wherein the link confidence represents a likelihood a first record from the first source and a second record from the second source represent a common entity.

### Claim 3 (depends on 1)

The method of claim 1, wherein the corpus is an internal corpus.

### Claim 4 (depends on 3)

The method of claim 3, wherein determining the one or more attributes is based on a second corpus and the second corpus is a public corpus.

### Claim 5 (depends on 1)

The method of claim 1, wherein the one or more attributes are selected from the group consisting of: a name, an address, a gender, an email address, a phone number, and an identification number.

### Claim 6 (depends on 1)

The method of claim 1 further comprising:
classifying each attribute of the one or more attributes.

### Claim 7 (depends on 1)

The method of claim 1, further comprising:
identifying, for at least one attribute of the one or more attributes, a language associated with the attribute.

### Claim 8 (depends on 1)

The method of claim 1, further comprising:
identifying, for at least one attribute of the one or more attributes, a regulation associated with the attribute.

### Claim 9 (depends on 8)

The method of claim 8, wherein the at least one attribute includes a location, and wherein the identifying the regulation is based on the location.

### Claim 10 (depends on 9)

The method of claim 9, wherein the regulation is a data protection regulation.

### Claim 11

A system configured to link records comprising:
a processor; and
a computer-readable storage medium communicatively coupled to the processor and storing program instructions which, when executed by the processor, are configured to cause the processor to:
identify a first source and a second source wherein the first source and the second source include one or more records and each record includes one or more attributes;
determine, based on a corpus, the one or more attributes;
generate, based on the attributes, a set of feature vectors, wherein the set of feature vectors represent the one or more attributes;
compare each record in the first source against each record in the second source;
generate, in response to the comparison, a link confidence;
link, in response to the link confidence being above a linking threshold, a first record from the first source and a second record from the second record;
determine a first feature vector of the set of feature vectors used in the linking; and
output a set of results.

### Claim 12 (depends on 11)

The system of claim 11, wherein the link confidence represents a likelihood a first record from the first source and a second record from the second source represent a common entity.

### Claim 13 (depends on 11)

The system of claim 11, wherein the corpus is an internal corpus.

### Claim 14 (depends on 13)

The system of claim 13, wherein the determination of the one or more attributes is based on a second corpus and the second corpus is a public corpus.

### Claim 15 (depends on 11)

The system of claim 11, wherein the program instructions further configured to cause the processor to:
classify each attribute of the one or more attributes.

### Claim 16

A computer program product, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processing unit to cause the processing unit to:
identify a first source and a second source wherein each of the sources include one or more records and each record includes one or more attributes;
determine, based on a corpus, the one or more attributes;
generate, based on the attributes, a set of feature vectors, wherein the set of feature vectors represent the one or more attributes;
compare each record in the first source against each record in the second source;
generate, in response to the comparison, a link confidence;
link, in response to the link confidence being above a linking threshold, a first record from the first source and a second record from the second record;
determine a first feature vector of the set of feature vectors used in the linking; and
output a set of results.

### Claim 17 (depends on 16)

The computer program product of claim 16, wherein the confidence represents a likelihood a first record from the first source and a second record from the second source represent a common entity.

### Claim 18 (depends on 16)

The computer program product of claim 16, wherein the corpus is an internal corpus.

### Claim 19 (depends on 18)

The computer program product of claim 18, wherein the determination of the one or more attributes is based on a second corpus and the second corpus is a public corpus.

### Claim 20 (depends on 16)

The computer program product of claim 16, wherein the program instructions further cause the processing unit to:
classify each attribute of the one or more attributes.