System, method, and computer program product to determine cuisine type classifiers

Patent metadata

Field	Value
Publication number	US-11334892-B2
Application number	US-201916667065-A
Country	US
Kind code	B2
Filing date	Oct 29, 2019
Priority date	Oct 29, 2018
Publication date	May 17, 2022
Grant date	May 17, 2022

Abstract

Provided is a method that includes determining cuisine type classifiers based on merchant data, receiving transaction data associated with each payment transaction of a plurality of payment transactions involving a plurality of merchants and a plurality of consumers; determining, with at least one processor, a plurality of cuisine type labels based on a plurality of merchant names associated with the plurality of payment transactions or a portion of the plurality of payment transactions, wherein a cuisine type label identifies a cuisine type provided by a respective merchant of the plurality of merchants; generating a merchant cuisine type model from the transaction data to determine a cuisine type label of a merchant, the merchant cuisine type model based on the transaction data, and makes an inference based on a combination of merchant or consumer features associated with a payment transaction of the plurality of payment transactions; and automatically predicting one or more cuisine type classifications using the merchant cuisine type model based on one or more payment transactions associated with a consumer. A system and a computer program product are also disclosed.

First claim

What is claimed is:

1. A computer-implemented method for classifying restaurant types by applying a label to each restaurant according to its cuisine type using only restaurant payment transactions processed within a closed payment network system, the method comprising:
receiving, by at least one processor associated with a cuisine type label generator, transaction data of a plurality of payment transactions involving a plurality of restaurant merchants and a plurality of consumers, wherein each of the consumers holds an account with a payment network and each of the restaurant merchants are in electronic communication with a payment network computing device using a merchant computing device, wherein each of the consumers initiates restaurant payment transactions on a consumer mobile computing device which is in electronic communication with a merchant computing device;
determining, with the cuisine type label generator, a plurality of cuisine type labels based on a plurality of restaurant names associated with the plurality of payment transactions or a portion of the plurality of payment transactions, wherein a cuisine type label identifies a cuisine type provided by a respective merchant of the plurality of restaurant merchants, and wherein the cuisine type labels are generated automatically based on (i) a predefined number of cuisine types filtered to include only restaurants popular in a predefined geographic region; (ii) a predefined number of seed words; and (iii) keywords extracted from restaurant names which are supported by comparison to labeled restaurants;
receiving, by a cuisine type model generator, cuisine type labels generated by the cuisine type label generator;
receiving, by the cuisine type model generator, from a restaurant feature extractor, restaurant operating features extracted from the transaction data, wherein the restaurant operating features comprise empirically generated correlations between restaurant operating features and cuisine types, including one or more of restaurant location, business hours, pricing and tipping culture, expected party size, and consumer visitation and loyalty patterns;
receiving, by the cuisine type model generator, from an embedding computing device, a plurality of embedding features comprising consumer embedded features and merchant restaurant embedded features, wherein the embedding features comprise correlations between consumer characteristics and restaurants as to cuisine types, including one or more of taste, cost, location, and efficiency, the merchant restaurant embedded features and the consumer embedded features automatically generated from transaction data, wherein each consumer feature of the consumer embedded features is used to characterize transaction data based on at least one of co-occurrence, frequency, or context of one or more words in a payment transaction of a consumer, and the payment transaction is associated with a purchase at a merchant, and wherein one or more merchant features of the merchant restaurant embedded features include each consumer feature that is associated with a restaurant merchant from which the payment transaction of the consumer originated;
outputting, by the cuisine type model generator, to a restaurant cuisine type model, the restaurant operating features, consumer embedded features and merchant restaurant embedded features,
wherein the restaurant cuisine type model combines the restaurant operating features with the consumer embedded features and merchant restaurant embedded features in a plurality of parallel input paths using a deep learning neural network to predict cuisine type labels, based on the labels generated by the cuisine type label generator for restaurant merchants associated with the transaction data;
training the restaurant cuisine type model, by the cuisine type model generator, using labeled restaurant names;
receiving, by the restaurant cuisine type model, in payment transaction data from the restaurant merchant computing device or the payment network computing device, a payment transaction initiated by the consumer mobile computing device;
automatically predicting, by the restaurant cuisine type model, one or more cuisine type labels using the restaurant cuisine type model based on the payment transaction associated with the consumer; and
applying one or more of the cuisine type labels to the restaurant merchant and associating the cuisine type labels with the consumer.

2. The computer-implemented method of claim 1 , further comprising:
generating, with at least one processor, a recommendation based on the one or more cuisine type labels.

3. The computer-implemented method of claim 1 , further comprising:
detecting, with at least one processor, a payment fraud based at least partially on the one or more cuisine type labels.

4. The computer-implemented method of claim 1 , wherein the transaction data associated with the payment transaction comprises at least one of:
data associated with an identification of the consumer controlling an account associated with the payment transaction;
data associated with the merchant in the payment transaction;
data associated with a location of the merchant in the payment transaction;
data associated with a time interval at which the payment transaction occurred;
data associated with an authorized transaction amount of the payment transaction; or
data associated with a settlement transaction amount of the payment transaction.

5. The computer-implemented method of claim 1 , further comprising:
determining a cuisine type label of the plurality of cuisine type labels for at least one merchant for the payment transaction of the consumer based on one or more characteristics in the payment transaction associated with the at least one merchant and the consumer.

6. The computer-implemented method of claim 5 , further comprising:
determining merchant attributes by encoding a plurality of consumer embedded features of a respective plurality of consumers of the merchant;
determining consumer characteristics by encoding a plurality of restaurant merchant embedded features for the consumer; and
determining compatibility between the merchant and the consumer based on the similarity of characteristics of the consumer to the merchant attributes.

7. The computer-implemented method of claim 6 , wherein the merchant or consumer include business features comprising at least one of:
an authorized amount percentile per payment transaction for the merchant;
a tip percentile per payment transaction for the merchant;
an hourly transaction percentile based on a number of payment transactions conducted in an hourly time interval for the merchant;
a party size distribution per payment transaction for the merchant;
a mean cost per payment transaction based on expected party size for the merchant;
a distribution of a number of payment transactions conducted based on days-of-week for the merchant;
a distribution of a number of transactions conducted based on hours-of-day during a weekday for the merchant; or
a distribution of a number of transactions conducted based on hours-of-day during a weekend day for the merchant.

8. The computer-implemented method of claim 1 , wherein determining the plurality of cuisine type labels comprises:
determining a label of a merchant name associated with the payment transaction by satisfying criteria including at least one of support, confidence, or a significance of a word within the merchant name of the plurality of restaurant merchants.

9. The computer-implemented method of claim 1 , predicting the one or more cuisine type classifications using the restaurant cuisine type model, further comprises:
determining a plurality of merchant featur

Assignees

• Visa Int Service Ass

Inventors

• Hamooni Hossein
• Dev Himel

Classifications

• G06Q20/145

  Payments according to the detected use or quantity · CPC title

• G06Q20/4014 Primary

  Identity check for transactions · CPC title

• G06Q20/0855

  involving a third party · CPC title

• G06Q50/12 Primary

  Hotels or restaurants · CPC title

• G06Q20/4016

  involving fraud or risk level assessment in transaction processing · CPC title