# US 2020/0402015 A1 · Contextual search ranking using entity topic representations

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

- Number: US 2020/0402015 A1
- Document ID: US20200402015A1
- Status: Published application
- Kind: A1
- Application number: 16/446,364
- Filed: 2019-06-19
- Published: 2020-12-24
- Earliest claimed: 2019-06-19
- Assignees: Microsoft Technology Licensing, LLC
- Applicants: Microsoft Technology Licensing, LLC
- Inventors: Cagri Ozcaglar; Gio Carlo C. Borje; Sahin C. Geyik; Gurwinder S. Gulati; Ketan Thakkar
- Web page: https://patentgrep.com/patent/US20200402015A1

## CPC

- G06Q 10/1053: Employment or hiring
- G06F 16/24578: {using ranking}
- G06F 16/9038: Presentation of query results
- G06F 16/906: Clustering; Classification
- G06N 20/00: Machine learning
- G06Q: INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS

## Abstract

The disclosed embodiments provide a system for processing data. During operation, the system obtains a first embedding generated by a topic model from parameters of searches by a first recruiting entity and obtains a set of additional embeddings generated by the topic model from attributes of a set of candidates. Next, the system determines match features that include measures of similarity between the first embedding and each embedding in the set of additional embeddings. The system then applies a machine learning model to the match features and additional features for the candidates to produce a set of scores for the candidates. Finally, the system generates a ranking of the candidates according to the scores and outputs at least a portion of the ranking as search results of a current search by the first recruiting entity.

## Claims

### Claim 1

A method, comprising:
obtaining a first embedding generated by a topic model from parameters of one or more searches by a first recruiting entity;
obtaining a set of additional embeddings generated by the topic model from attributes of a set of candidates;
determining, by one or more computer systems, match features comprising measures of similarity between the first embedding and each embedding in the set of additional embeddings;
applying, by the one or more computer systems, a machine learning model to the match features and additional features for the set of candidates to produce a set of scores for the set of candidates;
generating a ranking of the set of candidates according to the set of scores; and
outputting at least a portion of the ranking as search results of a current search by the first recruiting entity.

### Claim 2 (depends on 1)

The method of claim 1, further comprising:
generating the topic model based on bag-of-words representations of documents associated with additional searches by recruiting entities and the attributes of additional candidates.

### Claim 3 (depends on 1)

The method of claim 1, further comprising:
obtaining a second embedding generated by the topic model from additional parameters of one or more additional searches by a second recruiting entity;
determining additional match features comprising the measures of similarity between the second embedding and each embedding in the set of additional embeddings; and
applying the machine learning model to the additional match features to produce the set of scores.

### Claim 4 (depends on 3)

The method of claim 3, wherein the first recruiting entity and the second recruiting entity comprise a recruiter and a recruiting contract.

### Claim 5 (depends on 1)

The method of claim 1, wherein obtaining the first embedding comprises:
generating a bag-of-words representation of the parameters from the one or more searches by the first recruiting entity over a period; and
applying the topic model to the bag-of-words representation to generate the first embedding.

### Claim 6 (depends on 1)

The method of claim 1, wherein obtaining the first embedding comprises:
applying a first topic model to a first subset of parameters of the one or more searches to generate a first sub-embedding;
applying a second topic model to a second subset of parameters of the one or more searches to generate a second sub-embedding; and
combining the first and second sub-embeddings into the first embedding.

### Claim 7 (depends on 1)

The method of claim 1, wherein obtaining the set of additional embeddings comprises:
identifying the set of candidates that match one or more parameters of the current search; and
applying the topic model to the attributes of the set of candidates to generate the set of additional embeddings.

### Claim 8 (depends on 1)

The method of claim 1, wherein determining the match features comprising the measures of similarity between the first embedding and each embedding in the set of additional embeddings comprises:
calculating a first match feature as a cosine similarity between the first embedding and each embedding in the set of additional embeddings; and
calculating a second match feature as a Jensen-Shannon divergence between the first embedding and each embedding in the set of additional embeddings.

### Claim 9 (depends on 1)

The method of claim 1, wherein the additional features comprise at least one of:
an exact match between an attribute of a candidate and one or more parameters of the current search;
a reputation score for a skill of the candidate; and
a job-seeking status of the candidate.

### Claim 10 (depends on 1)

The method of claim 1, wherein the first embedding and the set of additional embeddings comprise a distribution of topics in a document.

### Claim 11 (depends on 1)

The method of claim 1, wherein the attributes of the set of candidates comprise at least one of:
a skill;
a title;
a location;
an industry;
a seniority;
a company; and
an educational background.

### Claim 12

A system, comprising:
one or more processors; and
memory storing instructions that, when executed by the one or more processors, cause the system to:
apply a topic model to parameters of one or more searches by a first recruiting entity to produce a first embedding;
apply the topic model to attributes of a set of candidates to produce a set of additional embeddings;
determine match features comprising measures of similarity between the first embedding and each embedding in the set of additional embeddings; and
output the match features for use in generating search results of a current search by the first recruiting entity.

### Claim 13 (depends on 12)

The system of claim 12, wherein the memory further stores instructions that, when executed by the one or more processors, cause the system to:
apply the topic model to additional parameters of one or more additional searches by a second recruiting entity to produce a second embedding; and
determine additional match features comprising the measures of similarity between the second embedding and each embedding in the set of additional embeddings.

### Claim 14 (depends on 13)

The system of claim 13, wherein the first recruiting entity and the second recruiting entity comprise a recruiter and a recruiting contract.

### Claim 15 (depends on 12)

The system of claim 12, wherein outputting the match features for use in generating search results of the current search by the first recruiting entity comprises:
inputting the match features with additional features of the set of candidates into a machine learning model;
receiving, as output from the machine learning model, a set of scores for the set of candidates;
generating a ranking of the set of candidates according to the set of scores; and
outputting at least a portion of the ranking as search results of the current search.

### Claim 16 (depends on 12)

The system of claim 12, wherein the memory further stores instructions that, when executed by the one or more processors, cause the system to:
identify the set of candidates that match one or more parameters of the current search.

### Claim 17 (depends on 12)

The system of claim 12, wherein determining the match features comprising the measures of similarity between the first embedding and each embedding in the set of additional embeddings comprises:
calculating a first match feature as a cosine similarity between the first embedding and each embedding in the set of additional embeddings; and
calculating a second match feature as a Jensen-Shannon divergence between the first embedding and each embedding in the set of additional embeddings.

### Claim 18 (depends on 12)

The system of claim 12, wherein the first embedding and the set of additional embeddings comprise a distribution of topics in a document.

### Claim 19

A non-transitory computer-readable storage medium storing instructions that when executed by a computer cause the computer to perform a method, the method comprising:
obtaining a first embedding generated by a topic model from parameters of one or more searches by a first recruiting entity;
obtaining a set of additional embeddings generated by the topic model from attributes of a set of candidates;
determining match features comprising measures of similarity between the first embedding and each embedding in the set of additional embeddings;
applying, by the one or more computer systems, a machine learning model to the match features and additional features for the set of candidates to produce a set of scores for the set of candidates;
generating a ranking of the set of candidates according to the set of scores; and
outputting at least a portion of the ranking as search results of a current search by the first recruiting entity.

### Claim 20 (depends on 19)

The non-transitory computer readable storage medium of claim 19, wherein determining the match features comprising the measures of similarity between the first embedding and each embedding in the set of additional embeddings comprises:
calculating a first match feature as a cosine similarity between the first embedding and each embedding in the set of additional embeddings; and
calculating a second match feature as a Jensen-Shannon divergence between the first embedding and each embedding in the set of additional embeddings.