Recommendation system using improved neural network

Patent metadata

Field	Value
Publication number	US-10896459-B1
Application number	US-202016842432-A
Country	US
Kind code	B1
Filing date	Apr 7, 2020
Priority date	Nov 28, 2016
Publication date	Jan 19, 2021
Grant date	Jan 19, 2021

Abstract

Some aspects of the present disclosure relate to generating and training a neural network by separating historical item interaction data into both inputs and outputs. This may be done, for example, based on date. For example, a neural network machine learning technique may be used to generate a prediction model using a set of inputs that includes both a number of items purchased by a number of users before a certain date as well as some or all attributes of those items, and a set of outputs that includes the items purchased after that date. The items purchased before that date and the associated attributes can be subjected to a time-decay function.

First claim

What is claimed is:

1. A method, comprising:
by execution of program instructions by a computing system that comprises one or more computing devices:
maintaining user-specific event histories for each of a plurality of users of an electronic catalog, each event history corresponding to a respective user and comprising event data representing user-item interaction events performed by the user, wherein the event data for each user-item interaction event includes an identifier of a catalog item corresponding to the event and a time of the user-item interaction event;
dividing each user-specific event history such that user-item interaction events that occurred before a split time are allocated exclusively to an input group and user-item interaction events that occurred after the split time are allocated exclusively to an expected output group; and
training a neural network to predict user-item interaction events falling in the expected output group, wherein training the neural network comprises (1) providing, to input layer nodes of the neural network, identifiers of catalog items represented in the input group, and (2) providing, to output layer nodes of the neural network, identifiers of catalog items represented in the expected output group.

2. The method of claim 1 , further comprising generating, with the trained neural network, a score representing a likelihood that a user that interacts with a particular first catalog item will also interact with a particular second catalog item.

3. The method of claim 2 , further comprising selecting the second catalog item to recommend to the user based on the score, and in response to selecting the second catalog item to recommend, generating, and outputting for display to a device of the user, an electronic page that includes a representation of the second catalog item.

4. The method of claim 1 , wherein the input group corresponds to a time window that is substantially shorter in duration than a time window corresponding to the expected output group.

5. The method of claim 1 , wherein training the neural network further comprises providing, to the input layer nodes, attributes of catalog items represented in the input group.

6. The method of claim 5 , wherein the attributes of the catalog items include item brands.

7. The method of claim 1 , wherein training the neural network further comprises providing, to the input layer nodes, user attributes of users corresponding to the user-specific event histories.

8. The method of claim 7 , wherein the user attributes include geographic locations of users.

9. The method of claim 1 , wherein the user-item interaction events include item purchase events.

10. A system, comprising:
a data repository that stores user-specific event histories for each of a plurality of users of an electronic catalog, each event history corresponding to a respective user and comprising event data representing user-item interaction events performed by the user, wherein the event data for each user-item interaction event includes an identifier of a catalog item corresponding to the event and a time of the user-item interaction event; and
a computing system comprising one or more computing devices, the computing system programmed with executable program instructions to train a neural network by a process that comprises:
dividing the user-specific event histories such that user-item interaction events that occurred before a split time are allocated exclusively to an input group and user-item interaction events that occurred after the split time are allocated exclusively to an expected output group; and
training the neural network to predict user-item interaction events falling in the expected output group, wherein training the neural network comprises (1) providing, to input layer nodes of the neural network, identifiers of catalog items represented in the input group, and (2) providing, to output layer nodes of the neural network, identifiers of catalog items represented in the expected output group.

11. The system of claim 10 , wherein the computing system is configured to implement a prediction model that uses the trained neural network to generate scores representing likelihoods that particular users will interact with particular catalog items.

12. The system of claim 11 , wherein the computing system is further configured to implement a recommendation engine that selects catalog items to recommend to users based on the scores.

13. The system of claim 10 , wherein the input group corresponds to a time window that is substantially shorter in duration than a time window corresponding to the expected output group.

14. The system of claim 10 , wherein the computing system, in training the neural network, is configured to provide, to the input layer nodes, attributes of catalog items represented in the input group.

15. The system of claim 14 , wherein the attributes of the catalog items include item brands.

16. The system of claim 10 , wherein the computing system, in training the neural network, is configured to provide, to the input layer nodes, user attributes of users corresponding to the user-specific event histories.

17. The system of claim 16 , wherein the user attributes include geographic locations of users.

18. Non-transitory computer storage that stores program instructions that direct a computing system to perform a process that comprises:
maintaining user-specific event histories for each of a plurality of users of an electronic catalog, each event history corresponding to a respective user and comprising event data representing user-item interaction events performed by the user, wherein the event data for each user-item interaction event includes an identifier of a catalog item corresponding to the event and a time of the user-item interaction event;
dividing the user-specific event histories such that user-item interaction events that occurred before a split time are allocated exclusively to an input group and user-item interaction events that occurred after the split time are allocated exclusively to an expected output group; and
training a neural network to predict user-item interaction events falling in the expected output group, wherein training the neural network comprises (1) providing, to input layer nodes of the neural network, identifiers of catalog items represented in the input group, and (2) providing, to output layer nodes of the neural network, identifiers of catalog items represented in the expected output group.

19. The non-transitory computer storage of claim 18 , wherein the program instructions further direct the computing system to generate catalog item recommendations for users based on scores generated by the trained neural network.

Assignees

• Amazon Tech Inc

Inventors

• Joseph Rejith George
• Rybakov Oleg

Classifications

• G06N3/08

  Learning methods · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title

• G06N3/0499

  Feedforward networks · CPC title

• G06N3/0495

  Quantised networks; Sparse networks; Compressed networks · CPC title