Recommendation engine comprising an inference module for associating users, households, user groups, product metadata and transaction data and generating aggregated graphs using clustering

Patent metadata

Field	Value
Publication number	US-10445811-B2
Application number	US-201514610521-A
Country	US
Kind code	B2
Filing date	Jan 30, 2015
Priority date	Oct 27, 2014
Publication date	Oct 15, 2019
Grant date	Oct 15, 2019

Abstract

A method for providing recommendations for customers is described. The method comprises obtaining transaction data associated with one or more products being purchased by a customer. The method further comprises generating a customer household graph, wherein the customer household graph comprises a customer household node associated with the customer, a product superset node, and one or more product nodes associated with the product superset node. Further, at least one product metadata node associated with each of the one or more product nodes is determined. Further, at least one user node is associated with the customer household node based on the at least one product metadata node. Further, a user group node associated with the customer household node is determined based on the at least one user node. Further, one or more recommendations to be provided to the user are ascertained.

First claim

We claim:

1. A recommendation engine comprising:
a processor;
a data acquisition module coupled to the processor to,
obtain transaction data associated with one or more products being purchased by a customer in real-time, wherein the transaction data is represented as a graph database layer over a repository, and at the graph database layer an object and relations associated with the object are stored as nodes and edges and a write operation and a read operation fetches the transaction data from the repository;
generate a customer household graph based on the transaction data, wherein the customer household graph comprises a customer household node associated with the customer, a product superset node associated with the customer household node by ascertaining one or more relations, and one or more product nodes associated with the product superset node, wherein each of the one or more product nodes represents a product from amongst the one or more products; and

an inference module coupled to the processor to,
determine at least one product metadata node associated with each of the one or more product nodes based on the transaction data by ascertaining a relation between each of the one or more product nodes and corresponding at least one product metadata node and compute a confidence value associated with the at least one product metadata, wherein the computed confidence value indicates accuracy of identification of the product metadata node, wherein the at least one product metadata node represents product metadata associated with the product;
associate at least one user node with the customer household node by ascertaining the at least one user node based on the at least one product metadata node and the transaction data, wherein the at least one user node corresponds to the at least one product metadata, wherein the at least one user node is generated along with another confidence value, based on a predefined rule set or a training set, wherein at least one probable user of a product is represented and subsequently associated with the customer household node as the at least one user node;
determine a user group node associated with the customer household node based on the at least one user node, wherein the user group node indicates a user group comprising one or more users associated with the customer household node; and

a recommendation module coupled to the processor to ascertain one or more recommendations to be provided to the customer based on at least one of the user node, the user group node, a cluster of the user group nodes, and a recommendation training dataset,
wherein the inference module stores the customer household graph and product metadata nodes as an inference graph, and also stores the inference graph, user nodes, user graph nodes as an aggregated graph,
wherein the recommendation engine further comprises:
a clustering module coupled to the processor to perform similarity measure of determined user group node and accordingly cluster the user group node into at least one cluster when the similarity measure is above a predefined threshold.

2. The recommendation engine as claimed in claim 1 , wherein the data acquisition module further is to,
generate a spatial node indicative of a location of the purchase, wherein the spatial node represents spatial data obtained in real-time; and
generate a temporal node indicative of a time of the purchase, wherein the temporal node represents temporal data obtained in real-time.

3. The recommendation engine as claimed in claim 1 , wherein the inference module ascertains the at least one probable user by performing a first inference based on the at least one product metadata node and the transaction data.

4. A method for providing recommendations for customers, the method comprising:
obtaining transaction data associated with one or more products being purchased by a customer in real-time, wherein the transaction data is represented as a graph database layer over a repository, and at the graph database layer an object and relations associated with the object are stored as nodes and edges and a write operation and a read operation fetches the transaction data from the repository;
generating a customer household graph based on the transaction data, wherein the customer household graph comprises a customer household node associated with the customer, a product superset node, and one or more product nodes associated with the product superset node by ascertaining one or more relations between each of the one or more product nodes and the product superset node, wherein each of the one or more product nodes represents a product from amongst the one or more products;
determining at least one product metadata node associated with each of the one or more product nodes based on the transaction data by ascertaining a relation between each of the one or more product nodes and corresponding at least one product metadata node and computing a confidence value associated with the at least one product metadata node, wherein the computed confidence value indicates accuracy of identification of the product metadata node, wherein the at least one product metadata node represents product metadata associated with the product;
associating at least one user node with the customer household node by ascertaining the at least one user node based on the at least one product metadata node and the transaction data, wherein the at least one user node corresponds to the at least one product metadata node, wherein the at least one user node is generated along with another confidence value, based on a predefined rule set or a training set, wherein at least one probable user of a product is represented and subsequently associated with the customer household node as the at least one user node;
determining a user group node associated with the customer household node based on the at least one user node, wherein the user group node indicates a user group comprising one or more users associated with the customer household node; and
ascertaining one or more recommendations to be provided to the user,
wherein the customer household graph and product metadata nodes are stored as an inference graph, and the inference graph, user nodes, user graph nodes are stored as an aggregated graph,
wherein the method further comprises:
performing similarity measure of the determined user group node and accordingly cluster the user group node into at least one cluster when the similarity measure is above a predefined threshold.

5. The method as claimed in claim 4 , wherein the transaction data comprises product data, product metadata, spatial data, and temporal data, wherein the product data and the product metadata comprises information associated with the one or more products, and wherein the spatial data comprises information indicative of a location where the customer is purchasing the product, and wherein the temporal data comprises information indicative of a date and a time of purchase of the one or more products.

6. The method as claimed in claim 4 , wherein the associating further comprises performing a first inference based on the at least one product metadata node and the transaction data for obtaining the at least one user node.

7. The method as claimed in claim 4 , wherein the one or more recommendations are ascertained based on at least one of the user node, the user group node, the transaction data, one or more clusters, and a recommendation training dataset.

8. The method as claimed in claim 4 , wherein the another confidence value is associated with the at least one user node, wherein the another confidence value indicates accuracy of identification of the user.

9. The method as claimed in claim 4 , w

Assignees

• Tata Consultancy Services Ltd

Inventors

• Chacko Viju
• Ramaswamy Satyanarayanan
• Sarkar Shampa

Classifications

• G06Q30/0201

  Market modelling; Market analysis; Collecting market data · CPC title

• G06Q30/0631 Primary

  Recommending goods or services · CPC title