Machine learning techniques to predict geographic talent flow

Patent metadata

Field	Value
Publication number	US-11238352-B2
Application number	US-201815941236-A
Country	US
Kind code	B2
Filing date	Mar 30, 2018
Priority date	Mar 30, 2018
Publication date	Feb 1, 2022
Grant date	Feb 1, 2022

Abstract

Techniques are provided for predicting talent flow to and/or from a geographical region. In one technique, multiple entity profiles are stored and analyzed to generate training data that is labeled indicating whether a corresponding entity has moved to or moved from a region. A machine-learned prediction model is generated or trained based on the training data. Using the machine-learned prediction model, a prediction is made whether, for each entity corresponding to another entity profile, that entity will move to or move from a particular geographic region. Based on multiple predictions, a number of entities that are predicted to move to or move from the particular geographic region is determined. Talent flow data that is based on the number of entities is presented on a computer display.

First claim

What is claimed is:

1. A method comprising:
storing a plurality of entity profiles;
generating training data based on a first subset of the plurality of entity profiles, wherein at least a portion of the training data is labeled indicating whether a corresponding entity has moved to or moved from a region;
generating a machine-learned prediction model based on the training data;
using the machine-learned prediction model to predict whether, for each entity corresponding to an entity profile in a second subset of the plurality of entity profiles, said each entity will move to or move from a particular region;
based on a plurality of predictions for entities corresponding to the second subset of entity profiles, determining a number of entities who are predicted to move to or move from the particular region;
for each entity that is predicted to move to or move from the particular region:
analyzing an entity profile of said each entity to identify one or more skills that are associated with said each entity; and
for each skill of a plurality of skills, determining a total number of instances, in the multiple entity profiles, of said each skill; and

causing talent flow data, based on the number of entities, and skill data, based on the total number of instances of at least one skill in the plurality of skills, to be presented on a computer display;
wherein the method is performed by one or more computing devices.

2. The method of claim 1 , wherein the number of entities are predicted to move to the particular region, the method further comprising:
using the machine-learned prediction model to predict whether, for each entity corresponding to an entity profile in the second subset of the plurality of entity profiles, said each entity will move from the particular region;
based on a second plurality of predictions for entities corresponding to the second subset of entity profiles, determining a second number of entities who are predicted to move from the particular region;
causing second talent flow data that is based on the second number of entities to be presented on the computer display.

3. The method of claim 2 , further comprising:
for each entity that is predicted to move from the particular region:
analyzing an entity profile of said each entity to identify one or more skills that are associated with said each entity; and
for each skill of a plurality of skills, determining a total number of instances, in the multiple entity profiles, of said each skill; and
for a particular skill in the plurality of skills, determining a net skill flow for the particular skill;
wherein causing the skill data to be presented on the computer display comprises causing the net skill flow for the particular skill to be presented on the computer display.

4. The method of claim 1 , wherein the number of entities are predicted to move from the particular region, the method further comprising:
based on the plurality of predictions:
determining a first number of entities that are predicted to move to a first region that is different than the particular region, wherein the first number of entities is a first subset of the number of entities;
determining a second number of entities that are predicted to move to a second region that is different than the first region and the particular region,
wherein the second number of entities is a second subset of the number of entities.

5. The method of claim 1 , wherein the number of entities are predicted to move to the particular region, the method further comprising:
based on the plurality of predictions:
determining a first number of entities that are predicted to move from a first region that is different than the particular region, wherein the first number of entities is a first subset of the number of entities;
determining a second number of entities that are predicted to move from a second region that is different than the first region and the particular region,
wherein the second number of entities is a second subset of the number of entities.

6. The method of claim 1 , wherein the machine-learned prediction model is based on a plurality of features, wherein the plurality of features includes one or more of: a number of organizations, in a certain region, that an entity has followed; a number of organizations, in the certain region, that the entity has liked; or a number of connections that the entity has established with other entities associated with the certain region.

7. The method of claim 1 , wherein the machine-learned prediction model is based on a plurality of features, wherein the plurality of features includes one or more of: whether or a number of electronic messages that an entity has sent to employees of an organization in a certain region; whether or a number of requests by the entity of profiles of organizations in the certain region; whether or a number of requests by the entity of job postings associated with the certain region; whether or a number of applications by the entity to job openings offered by one or more organizations in the certain region; or whether or a number of requests by the entity to view profiles of other entities affiliated with one or more organizations in the certain region.

8. The method of claim 1 , wherein the machine-learned prediction model is based on a plurality of features, wherein the plurality of features includes one or more of: whether or a number of times an entity has initiated a search and selected a search result that is associated with a certain region; the entity adding one or more skills to a profile of the entity; or the entity receiving an endorsement from another entity.

9. The method of claim 1 , wherein the machine-learned prediction model is based on a plurality of user-related features and a plurality of region-centric features.

10. The method of claim 9 , wherein the plurality of region-centric features includes one or more of: a number of job openings in a certain region; a cost of living in the certain region; an amount, level or rate of taxes associated with the certain region; a quality of public schools in the certain region; a quality of public utilities in the certain region; or a cost of public utilities in the certain region.

11. A method comprising:
storing a plurality of entity profiles;
based on a particular period of time:
analyzing a first subset of the plurality of entity profiles to determine a first number of entities that have moved to a particular region during the particular period of time;
analyzing a second subset of the plurality of entity profiles to determine a second number of entities that have moved from the particular region during the particular period of time;
causing data about the first number of entities and the second number of entities to be presented on a computer display;
for each entity corresponding to the first number of entities:
using an entity profile corresponding to said each entity to identify one or more first skills of said entity;
updating a skill inflow counter for each skill of the one or more first skills;
generating first skill data based on the skill inflow counter for each skill of the one or more first skills;
causing the first skill data to be presented on the computer display;
for each entity corresponding to the second number of entities:
using an entity profile corresponding to said each entity to identify one or more second skills of said entity;
updating a skill outflow counter for each skill of the one or more second skills;
generating second skill data based on the skill outflow counter for each skill of the one or more second skills;
causing the second skill data to be presented on the com

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Kuan Chi-Yi
• Huang Shen
• Wang Yu
• Zhang Yongzheng
• Ko Paul
• Elasra Shady
• Bu Fanbin

Classifications

• G06Q10/063 Primary

  Operations research, analysis or management · CPC title

• G06N20/00

  Machine learning · CPC title

• G06Q10/04

  Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" (market predictions or forecasting for commercial activities G06Q30/0202) · CPC title

• G06N5/04 Primary

  Inference or reasoning models · CPC title

• H04L67/306

  User profiles · CPC title