Techniques for identifying similar products

Patent metadata

Field	Value
Publication number	US-10339586-B1
Application number	US-201313974877-A
Country	US
Kind code	B1
Filing date	Aug 23, 2013
Priority date	Aug 23, 2013
Publication date	Jul 2, 2019
Grant date	Jul 2, 2019

Abstract

Methods and apparatus are described for identifying similar products or services for the purpose of making relevant recommendations to an online consumer. Products and services are represented by associated vectors which include values for each of a plurality of attributes of the corresponding product or service. One or more similar products or services are identified relative to a reference product or service set with reference to the distance between the end points of the respective vectors in the associated vector space.

First claim

What is claimed is:

1. A computer-implemented method for making product recommendations in a computing network, comprising:
identifying, using one or more computing devices operating in the network, a reference product set based in part on one or more actions of a user associated with a remote device, the reference product set including a first product and a second product, wherein the one or more actions include one or more of viewing information representing the first and/or second product on the remote device, selecting the first and/or second product on the remote device, or purchasing the first and/or second product;
transmitting, using the one or more computing devices, a first product interface control for presentation on the remote device;
receiving, using the one or more computing devices, first selection data representing activation of the first product interface control from the remote device;
retrieving, using the one or more computing devices, a first product vector associated with the first product from a data store in response to the first selection data, the first product vector comprising a first plurality of values corresponding to a plurality of product attributes, the plurality of product attributes defining a vector space, the first plurality of values defining a first point in the vector space;
retrieving, using the one or more computing devices, a second product vector associated with the second product from the data store in response to the first selection data, the second product vector comprising a second plurality of values corresponding to the plurality of product attributes, the second plurality of values defining a second point in the vector space;
identifying a third point in the vector space, the third point representing a third product having an associated third product vector, the third product vector comprising a third plurality of values corresponding to the plurality of product attributes, the third plurality of values defining the third point in the vector space, the third product not being in the reference product set;
determining, using the one or more computing devices, that the third product is similar to the reference product set by:
calculating a first Euclidean distance between the first point and the third point in the vector space;
calculating a second Euclidean distance between the second point and the third point in the vector space;
determining a degree to which the first product and the second product in the reference product set are similar by determining a third Euclidean distance between the first point and the second point;
determining, based on the third Euclidean distance, a programmable threshold Euclidean distance for the reference product set beyond which products not in the reference product set are not considered similar to the reference product set; and
determining a sum of the first Euclidean distance and the second Euclidean distance;
determining that the sum of the first Euclidean distance and the second Euclidean distance is within the programmable threshold Euclidean distance for the reference product set;

generating a detail page that includes the first product in the reference product set, the second product in the reference product set, and the third product; and
transmitting, using the one or more computing devices, information representing the detail page for presentation on the remote device, thereby indicating that the third product is a recommended product based on the reference product set.

2. The method of claim 1 , further comprising:
including the third product in the reference product set;
transmitting a second product interface control for presentation on the remote device;
receiving second selection data representing activation of the second product interface control from the remote device;
identifying a fourth product having an associated fourth product vector in response to the second selection data, the fourth product vector comprising a fourth plurality of values corresponding to the plurality of product attributes, the fourth plurality of values defining a fourth point in the vector space, wherein identifying the fourth product includes:
calculating a fourth distance between the first point and the fourth point in the vector space;
calculating a fifth distance between the second point and the fourth point in the vector space;
calculating a sixth distance between the third point and the fourth point in the vector space; and
identifying the fourth product based on the fourth distance, fifth distance, and the sixth distance.

3. The method of claim 1 , further comprising identifying the second product in the reference product set based on a distance between the first point in the vector space and the second point in the vector space.

4. A computer-implemented method for making product recommendations in a computing network, comprising:
identifying a first reference product set based in part on one or more actions of a user associated with a remote device operating in the network, the first reference product set including a first product and a second product, wherein:
the first product is associated with a first product vector, the first product vector comprising a first plurality of values corresponding to a plurality of product attributes, the plurality of product attributes defining a vector space, the first plurality of values defining a first point in the vector space; and
the second product is associated with a second product vector, the second product vector comprising a second plurality of values corresponding to the plurality of product attributes, the second plurality of values defining a second point in the vector space;

identifying a third point in the vector space, the third point representing a third product having an associated third product vector, the third product vector comprising a third plurality of values corresponding to the plurality of product attributes, the third plurality of values defining the third point in the vector space, the third product not being in the first reference product set;
determining that the third product is similar to the first reference product set by:
calculating a first distance between the first point and the third point in the vector space;
calculating a second distance between the second point and the third point in the vector space;
determining a degree to which the first product and the second product in the first reference product set are similar by determining a third distance between the first point and the second point;
determining, based on the third distance, a programmable threshold distance for the first reference product set beyond which products not in the first reference product set are not considered similar to the first reference product set; and
determining that a combination of the first distance and the second distance is within the programmable threshold distance for the first reference product set;

generating a detail page that includes the first product in the reference product set, the second product in the reference product set, and the third product; and
transmitting, information representing the detail page for presentation on the remote device, thereby indicating that the third product is a recommended product based on the first reference product set.

5. The method of claim 4 , further comprising retrieving the first product vector from a data store, the data store having a plurality of product vectors stored therein including the first and second product vectors, and identifying the second product in the reference product set by calculating a third distance between the first point in the vector space and the second point in the vector space.

6. The method of claim 4 , wherein the first plurality

Assignees

• Amazon Tech Inc

Inventors

• Khobragade Samir Bhayyaji
• Chen Qian

Classifications

• G06Q30/0631 Primary

  Recommending goods or services · CPC title