Search engine optimization by selective indexing

Patent metadata

Field	Value
Publication number	US-10846276-B2
Application number	US-201615131139-A
Country	US
Kind code	B2
Filing date	Apr 18, 2016
Priority date	Jun 30, 2015
Publication date	Nov 24, 2020
Grant date	Nov 24, 2020

Abstract

Example embodiments involve a system and methods for identifying valuable view item pages for search engine optimization. According to certain embodiments, the system performs operations that include predicting the probability of future traffic for a given product based on a number of product level factors as input variables, and identifying a selection of view item pages corresponding to the products with the probability of the highest future traffic in order to maximize the driving natural search traffic to a linked site of the corresponding view item page.

First claim

What is claimed is:

1. A system comprising:
one or more processors; and
a memory storing instructions that, when executed by at least one processor among the one or more processors, cause the system to perform operations comprising:
accessing a database that comprises input variables, the input variables comprising temporal data and including item level factors of a set of item listings for a product;
identifying a portion of the item level factors based at least in part on the temporal data;
generating a machine learned model data object based at least in part on the portion of the item level factors from the database;
assigning a target variable to the machine learned model data object, the target variable including a target search traffic value;
training the machine learned model data object to identify one or more of the item level factors that correspond with the target search traffic value;
calculating a search traffic value of an item listing for the product from among the set of item listings for the product based at least in part on the trained machine learned model data object, the search traffic value indicating a probability that the item listing for the product from among the set of item listings for the product is returned in response to user search;
determining that the search traffic value of the item listing for the product transgresses a threshold value;
indexing the item listing for the product based at least in part on assigning a metadata tag to the item listing for the product in response to the determining that the search traffic value transgresses the threshold value; and
returning the item listing for the product from among the set of item listings for the product to a search engine responsive to receiving a search request that identifies the product, based at least in part on the indexing of the item listing.

2. The system of claim 1 , wherein the machine learned model data object is a gradient boosted machine model.

3. The system of claim 1 , wherein the item level factors include one or more of:
a prior natural search traffic of the product;
a view count of the item listing and an item attribute of the item listing;
a watch count of the product;
a bounce count of the product; or
a quantity of the product sold.

4. The system of claim 1 , wherein the training the machine learned model data object is based at least in part on the target variable and the input variables and includes:
correlating a set of input variables from among the portion of the item level factors to the target variable.

5. The system of claim 1 , wherein the metadata tag is a hyper-text markup language.

6. The system of claim 1 , wherein the indexing the item listing further comprises:
assigning the metadata tag that indicates an index status assigned to the item listing for the product.

7. A method comprising:
accessing a database that comprises input variables, the input variables comprising temporal data and including item level factors of a set of item listings for a product;
identifying a portion of the item level factors based at least in part on the temporal data;
generating a machine learned model data object based at least in part on the portion of the item level factors from the database;
assigning a target variable to the machine learned model data object, the target variable including a target search traffic value;
training the machine learned model data object to identify one or more of the item level factors that correspond with the target search traffic value;
calculating a search traffic value of an item listing for the product from among the set of item listings for the product based at least in part on the trained machine learned model data object, the search traffic value indicating a probability that the item listing for the product from among the set of item listings for the product is returned in response to user search;
determining that the search traffic value of the item listing for the product transgresses a threshold value;
indexing the item listing for the product based at least in part on assigning a metadata tag to the item listing for the product in response to the determining that the search traffic value transgresses the threshold value; and
returning the item listing for the product from among the set of item listings for the product to a search engine responsive to receiving a search request that identifies the product, based at least in part on the indexing of the item listing.

8. The method of claim 7 , wherein the machine learned model data object is a gradient boosted machine model.

9. The method of claim 7 , wherein the item level factors include one or more of:
a prior natural search traffic of the product;
a view count of the item listing and an item attribute of the item listing;
a watch count of the product;
a bounce count of the product; or
a quantity of the product sold.

10. The method of claim 7 , wherein the training the machine learned model data object is based at least in part on the target variable and the input variables and includes:
correlating a set of input variables from among the portion of the item level factors to the target variable.

11. The method of claim 7 , wherein
the metadata tag is a hyper-text markup language tag.

12. A non-transitory machine-readable storage medium comprising instructions that, when executed by one or more processors of a machine, cause the machine to perform operations comprising:
accessing a database that comprises input variables, the input variables comprising temporal data and including item level factors of a set of item listings for a product;
identifying a portion of the item level factors based at least in part on at least the temporal data;
generating a machine learned model data object based at least in part on the portion of the input variables from the database;
assigning a target variable to the machine learned model data object, the target variable including a target search traffic value;
training the machine learned model data object to identify one or more of the item level factors that correspond with the target search traffic value;
calculating a search traffic value of an item listing for the product from among the set of item listings for the product based at least in part on the trained machine learned model data object, the search traffic value indicating a probability that the item listing for the product from among the set of item listings for the product is returned in response to user search;
determining that the search traffic value of the item listing for the product transgresses a threshold value;
indexing the item listing for the product based at least in part on assigning a metadata tag to the item listing for the product in response to the determining that the search traffic value transgresses the threshold value; and
returning the item listing for the product from among the set of item listings for the product to a search engine responsive to receiving a search request that identifies the product, based at least in part on the indexing of the item listing.

13. The non-transitory machine-readable storage medium of claim 12 , wherein the machine learned model data object is a gradient boosted machine model.

14. The non-transitory machine-readable storage medium of claim 12 , wherein the item level factors include one or more of:
a prior natural search traffic of the product;
a view count of the item listing and an item attribute of the item listing;
a watch count of the product;
a bounce count of the product; or
a quantity of the product sold.

Assignees

• Ebay Inc

Inventors

• Voleti Suman
• Talele Ganesh
• Islam Atiq
• Goel Ritu

Classifications

• G06N20/20

  Ensemble learning · CPC title

• G06Q30/0623

  by investigating goods or services · CPC title

• G06F16/958

  Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking · CPC title

• G06N20/00

  Machine learning · CPC title

• G06F16/2228 Primary

  Indexing structures · CPC title