Context-enhanced category classification to generate product category labels used to conduct electronic transactions based on regulatory restrictions

Patent metadata

Field	Value
Publication number	US-12248944-B2
Application number	US-202117547589-A
Country	US
Kind code	B2
Filing date	Dec 10, 2021
Priority date	Dec 10, 2021
Publication date	Mar 11, 2025
Grant date	Mar 11, 2025

Abstract

Systems/techniques for facilitating context-enhanced category classification are provided. In various embodiments, a system can access a first textual description of a product or service. In various aspects, the system can identify, via execution of named entity recognition, one or more keywords in the first textual description. In various instances, the system can access, from a set of queryable databases, one or more second textual descriptions that respectively correspond to the one or more keywords. In various cases, the system can generate, via execution of word embedding, a first numerical representation of the first textual description and one or more second numerical representations of the one or more second textual descriptions. In various aspects, the system can identify, via execution of a machine learning classifier, a category label for the product or service, based on the first numerical representation and the one or more second numerical representations.

First claim

What is claimed is:

1. A computer system, comprising:
computer-readable memory storing computer readable instructions;
a machine learning classifier, via which the computer system: (i) examines data, which contain electronic transactions regarding products and/or services, from a plurality of databases to which the machine learning classifier has been granted access, (ii) electronically applies, by an embedding component, a word embedding technique to the electronic transactions to generate a set of numerical representations; (iii) classifies, via a machine learning classifier, the numerical representations into category labels generated by a machine learning model that has been trained to perform named entity recognition and/or word embedding on data regarding the products or services that is provided as inputs to the machine learning model, and (iv) electronically determines if the category labels generated by the machine learning classifier based on outputs of the machine learning model are subject to one or more regulatory restrictions;
at least one processor that executes the computer-executable instructions stored in the computer-readable memory, which causes the computer system to perform operations comprising:
accessing a first textual description of a product or service, wherein the product or service is associated with an electronic transaction;
identifying, via execution of named entity recognition, one or more keywords in the first textual description;
accessing, from a set of queryable databases, one or more second textual descriptions that respectively correspond to the one or more keywords;
generating, via execution of word embedding, a first numerical representation of the first textual description and one or more second numerical representations of the one or more second textual descriptions;
identifying, via execution of a machine learning classifier, a category label for the product or service, based on the first numerical representation and the one or more second numerical representations; and
determining whether to permit the electronic transaction based on the category label.

2. The computer system of claim 1 , wherein the product or service is associated with an electronic transaction, and wherein the operations further comprise:
comparing the category label with one or more restricted categories; and
when the category label matches at least one of the one or more restricted categories, initiating a remedial action with respect to the electronic transaction.

3. The computer system of claim 2 , wherein the remedial action includes cancelling the electronic transaction.

4. The computer system of claim 2 , wherein the remedial action includes generating a restriction alert.

5. The computer system of claim 1 , wherein the machine learning classifier receives as input the first numerical representation concatenated with the one or more second numerical representations, and wherein the machine learning classifier produces as output the category label.

6. A computer system, comprising:
computer-readable memory storing computer readable instructions;
a machine learning classifier, via which the computer system: (i) examines data, which contain electronic transactions regarding products and/or services, from a plurality of databases to which the machine learning classifier has been granted access, (ii) electronically applies, by an embedding component, a word embedding technique to the electronic transactions to generate a set of numerical representations; (iii) classifies, via a machine learning classifier, the numerical representations into category labels generated by a machine learning model that has been trained to perform named entity recognition and/or word embedding on data regarding the products or services that is provided as inputs to the machine learning model, and (iv) electronically determines if the category labels generated by the machine learning classifier based on outputs of the machine learning model are subject to one or more regulatory restrictions;
at least one processor that executes the computer-executable instructions stored in the computer-readable memory, which causes the computer system to perform operations comprising:
accessing a first textual description of a product or service, wherein the product or service is associated with an electronic transaction;
identifying, via execution of the named entity recognition, one or more keywords in the first textual description;
accessing, from a set of queryable databases, one or more second textual descriptions that respectively correspond to the one or more keywords;
generating, via execution of the word embedding, a first numerical representation of the first textual description and one or more second numerical representations of the one or more second textual descriptions;
applying self-attention to the one or more second numerical representations, thereby yielding one or more adjusted numerical representations;
applying max pooling to the one or more adjusted numerical representations, thereby yielding an aggregated numerical representation;
identifying, via execution of a machine learning classifier, a category label for the product or service, based on the first numerical representation and the one or more second numerical representations, wherein the machine learning classifier receives as input the first numerical representation concatenated with the aggregated numerical representation, and wherein the machine learning classifier produces as output the category label;
determining whether to permit the electronic transaction based on the category label.

7. A computer system, comprising:
computer-readable memory storing computer readable instructions;
a machine learning classifier, via which the computer system: (i) examines data, which contain electronic transactions regarding products and/or services, from a plurality of databases to which the machine learning classifier has been granted access, (ii) electronically applies, by an embedding component, a word embedding technique to the electronic transactions to generate a set of numerical representations; (iii) classifies, via a machine learning classifier, the numerical representations into category labels generated by a machine learning model that has been trained to perform named entity recognition and/or word embedding on data regarding the products or services that is provided as inputs to the machine learning model, and (iv) electronically determines if the category labels generated by the machine learning classifier based on outputs of the machine learning model are subject to one or more regulatory restrictions;
at least one processor that executes the computer-executable instructions stored in the computer-readable memory, which causes the computer system to perform operations comprising:
accessing a training dataset, wherein the training dataset includes a set of concatenated inputs and a set of ground-truth category labels that respectively correspond to the set of concatenated inputs;
training the machine learning classifier via backpropagation on the training dataset;
accessing a first textual description of a product or service, wherein the product or service is associated with an electronic transaction;
identifying, via execution of the named entity recognition, one or more keywords in the first textual description;
accessing, from a set of queryable databases, one or more second textual descriptions that respectively correspond to the one or more keywords;
generating, via execution of the word embedding, a first numerical representation of the first textual description and one or more second numerical representations of the one or more second textual descriptions;
and identifying, via execu

Assignees

• Paypal Inc

Inventors

• Nguyen Van Hoang
• Gelli Francesco
• Chen Zhe
• Wang Hewen
• Tang Quan Jin Ferdinand
• Nahata Amit
• Upadhyay Rushik Navinbhai

Classifications

• G06F40/295

  Named entity recognition · CPC title

• G06F40/47

  Machine-assisted translation, e.g. using translation memory · CPC title

• G06F16/3334

  Selection or weighting of terms from queries, including natural language queries · CPC title

• G06Q20/407

  Cancellation of a transaction · CPC title

• G06Q30/06

  Buying, selling or leasing transactions · CPC title