- Number
- 12505110
- Published
- 2025-12-23
- Filed
- 2021-09-02
- Assignee
- Disney Enterprises, Inc.
- Inventors
- Batty, III; John Robert et al.
- CPC
- G06F16/24578
- Verdict
- Set aside matching/scoring algorithm, business
- Source
- Google Patents · FreePatentsOnline
Abstract
Techniques for dynamic matching include dynamically defining a first set of queries, including defining a plurality of weights for the first set of queries and defining one or more relationships between queries in the first set of queries. A first set of responses are received from a first participant in a first category of participants. A plurality of additional sets of responses are received from a second category of participants. A plurality of recommendation scores are generated, each corresponding to a potential pairing between the first participant and a participant in the second category of participants. Each respective recommendation score is generated based on: i) a comparison of the first set of responses and one of the additional sets of responses, ii) the defined plurality of weights, and iii) the defined one or more relationships. Matches are identified for the first participant based on the plurality of recommendation scores.
Background
BACKGROUND (1) Dynamic matching between different categories of participants is a challenging problem. Existing solutions typically identify a list of matching candidates for a given participant, and allow a participant to select from among the candidates in another category based on the participant's own criteria. But this is inefficient and burdensome for the participant. SUMMARY (2) Embodiments include a computer-implemented method. The method includes dynamically defining, using a computer processor, a first set of queries, including: defining a plurality of weights for the first set of queries, and defining one or more relationships between queries in the first set of queries. The method further includes receiving, at the computer processor, a first set of responses to the first set of queries from a first participant in a first category of participants. The method further includes receiving, at the computer processor, a plurality of additional sets of responses to the first set of queries, each additional set of responses being received from one of a plurality of participants in a second category of participants. The second category of participants is distinct from the first category of participants. The method further includes generating, using the computer processor, a plurality of recommendation scores, each respective recommendation score of the plurality of recommendation scores corresponding to a potential pairing between the first participant and one of the plura
Claims
1. A computer-implemented method comprising: training a machine learning model to generate queries using a training dataset comprising prior queries associated with an entity, a first category of participants, and a second category of participants; generating, by the trained machine learning model, a set of queries, wherein the set of queries is for a different entity, and wherein the set of queries is based on the first category of participants and the second category of participants; generating matching criteria associated with the set of queries, wherein the matching criteria relate to potential matches between participants associated with the different entity in the first category of participants and participants in the second category of participants; providing, to a particular participant, the set of queries generated by the trained machine learning model; defining a response time period for the particular participant to provide responses to the set of queries; receiving, from the particular participant, responses to the set of queries during the response time period, wherein the particular participant is associated with the first category of participants and the different entity; computing recommendation scores for the particular participant and multiple participants associated with the second category of participants and the different entity, wherein the recommendation scores are based on the matching criteria, the responses to the set of queries, and response data associated with the multiple participants; storing the computed recommendation scores in a first storage area configured for short-term storage during the response time period; modifying, during the response time period, at least one recommendation score of the computed recommendation scores in response to a change in the responses to the set of queries or a change in the response data associated with the multiple participants; restricting the particular participant from providing responses to the set of queries upon an expiration of the response time period; storing the computed recommendation scores in a second storage area configured for long-term storage upon the expiration of the response time period, wherein the first storage area configured for short-term storage during the response time period is associated with a lower computation expense than the second storage area configured for long-term storage; and generating a notification to the particular participant of the computed recommendation scores, wherein the notification includes indications of match strength for the multiple participants associated with the second category of participants. ||
12. A non-transitory computer-readable medium carrying instructions that, when executed by a computing system, cause the computing system to perform operations comprising: training a machine learning model to generate queries using a training dataset comprising prior queries associated with an entity, a first category of participants, and a second category of participants; generating, by the trained machine learning model, a set of queries the set of queries is for a different entity, and wherein the set of queries is based on the first category of participants and the second category of participants; generating matching criteria associated with the set of queries, wherein the matching criteria relate to potential matches between participants associated with a different entity in the first category of participants and participants in the second category of participants, providing, to a particular participant, the set of queries generated by the trained machine learning model; defining a response time period for the particular participant to provide responses to the set of queries; receiving, from the participant, responses to the set of queries during the response time period, wherein the particular participant is associated with the first category of participants and the different entity; computing recommendation scores for the particular participant and multiple participants associated with the second category of participants and the different entity, wherein the recommendation scores are based on the matching criteria, the responses to the set of queries, and response data associated with the multiple participants; restricting the particular participant from providing responses to the set of queries upon an expiration of the response time period; storing the computed recommendation scores in a first storage area configured for short-term storage during the response time period; modifying, during the response time period, at least one recommendation score of the computed recommendation scores in response to a change in the responses to the set of queries or a change in the response data associated with the multiple participants; storing the computed recommendation scores in a second storage area configured for long-term storage upon the expiration of the response time period, wherein the first storage area configured for short-term storage during the response time period is associated with a lower computation expense than the second storage area configured for long-term storage; and generating a notification to the particular participant of the computed recommendation scores, wherein the notification includes indications of match strength for the multiple participants associated with the second category of participants. ||
20. A system, comprising: at least one processor; and at least one non-transitory memory carrying instructions that, when executed by the at least one processor, cause the system to: train a machine learning model to generate queries using a training dataset comprising prior queries associated with an entity, a first category of participants, and a second category of participants; generate, by the trained machine learning model, a set of queries, wherein the set of queries is for a different entity, and wherein the set of queries is based on the first category of participants and the second category of participants; generate matching criteria associated with the set of queries, wherein the matching criteria relate to potential matches between participants associated with a different entity in the first category of participants and participants in the second category of participants; provide, to a particular participant, the set of queries generated by the trained machine learning model; define a response time period for the particular participant to provide responses to the set of queries; receive, from the particular participant, responses to the set of queries during the response time period, wherein the particular participant is associated with the first category of participants and the different entity; compute recommendation scores for the particular participant and multiple participants associated with the second category of participants and the different entity, wherein the recommendation scores are based on the matching criteria, the responses to the set of queries, and response data associated with the multiple participants; store the computed recommendation scores in a first storage area configured for short-term storage during the response time period; modify, during the response time period, at least one recommendation score of the computed recommendation scores in response to a change in the responses to the set of queries or a change in the response data associated with the multiple participants; restricting the particular participant from providing responses to the set of queries upon an expiration of the response time period; store the computed recommendation scores in a second storage area configured for long-term storage upon the expiration of the response time period, wherein the first storage area configured for short-term storage during the response time period is associated with a lower computation expense than the second storage area configured for long-term storage; and generate a notification to the particular participant of the computed recommendation scores, wherein the notification includes indications of match strength for the multiple participants associated with the second category of participants.