Refined search query results through external content aggregation and application

Patent metadata

Field	Value
Publication number	US-11301540-B1
Application number	US-201916351320-A
Country	US
Kind code	B1
Filing date	Mar 12, 2019
Priority date	Mar 12, 2019
Publication date	Apr 12, 2022
Grant date	Apr 12, 2022

Abstract

Systems and methods are disclosed for refining the accuracy of network searches by supplementing existing keywords and key phrases in an e-commerce catalog or other database with aggregated and analyzed additional, external data. The internet or another network can be crawled for identifiers which point to entries in the catalog or other database, and, subject to third-party use restrictions, data and metadata can be extracted to enrich the existing keywords and key phrases. The extracted external content may be processed by machine learning techniques in order to find similar entries in the original catalog or database. Categorizing and indexing the entries further improves search recall, including clustering via processing word embeddings.

First claim

What is claimed is:

1. A computer-implemented method, comprising:
clustering a plurality of offerings in an online marketplace, the clustering performed at least in part by a knowledge graph;
identifying, based on the clustering of the plurality of the offerings in the online marketplace, at least one subset of related offerings in the online marketplace;
accessing first content external to the online marketplace having an interface, the interface appearing on a web page including hypertext and further configured to receive a query;
performing at least one word embedding analysis on the first content to generate a context;
determining that at least one identifier in the first content relates to at least one entry in a search engine data store for the online marketplace, the at least one identifier being a hyperlink or a unique identifying code for a product in the online marketplace;
extracting at least one instance of data appearing in the first content, the extracted at least one instance of data including a keyword;
propagating the keyword to entries in the search engine data store for the at least one subset of related offerings; and
updating the search engine data store to reflect the context as well as an association between the extracted instance of data and the at least one entry.

2. The computer-implemented method of claim 1 , wherein the at least one identifier is a hyperlink, and a predetermined amount of text surrounding the hyperlink comprises the extracted at least one instance of data.

3. A computer-implemented method, comprising:
clustering a first set of entries in an online marketplace;
applying at least one similarity criterion to the first set of entries to identify an applicable subset, the at least one similarity criterion based at least in part on attributes of the first set of entries;
accessing first content external to a host device having a search engine and an interface, the interface configured to receive a query;
performing at least one word embedding analysis on the first content to generate a context;
determining that at least one identifier in the first content relates to at least one entry in the online marketplace;
extracting at least one instance of data appearing in the first content;
propagating the extracted instance of data to the subset; and
updating the online marketplace to reflect the context as well as an association between the extracted instance of data and the at least one entry.

4. The computer-implemented method of claim 3 , further comprising:
presenting query results on a client device, the query results based on the association between the extracted instance of data and the at least one entry.

5. The computer-implemented method of claim 3 , further comprising:
navigating a network using a web crawler, the network including a web page containing the first content.

6. The computer-implemented method of claim 3 , wherein the extracted at least one instance of data is a keyword or key phrase.

7. The computer-implemented method of claim 6 , further comprising:
locating the keyword or key phrase in the first content; and
matching the keyword or key phrase with product-identifying metadata in the first content.

8. The computer-implemented method of claim 7 , further comprising:
training a neural network on a sample data set; and
applying the neural network to optimize determination of the association between the keyword or key phrase and the at least one entry.

9. The computer-implemented method of claim 3 , wherein the at least one identifier is one of a unique character-based identifier or a hyperlink.

10. The computer-implemented method of claim 3 , further comprising:
indexing, in the online marketplace, the at least one identifier and the extracted at least one instance of data.

11. The computer-implemented method of claim 3 , wherein the clustering is performed at least in part by a knowledge graph.

12. A system, comprising:
at least one processor; and
a memory device including instructions that, when executed by the at least one processor, cause the system to:
cluster a first set of entries in an online marketplace;
apply at least one similarity criterion to the first set of entries to identify an applicable subset, the at least one similarity criterion based at least in part on attributes of the first set of entries;
access first content external to a host device having a search engine and an interface, the interface configured to receive a query;
perform at least one word embedding analysis on the first content to generate a context;
determine that at least one identifier in the first content relates to at least one entry in the online marketplace;
extract at least one instance of data appearing in the first content;
propagate the extracted instance of data to the subset; and
update the online marketplace to reflect the context as well as an association between the extracted instance of data and the at least one entry.

13. The system of claim 12 , wherein the instructions, when executed by the at least one processor, further cause the system to:
navigate a network using a web crawler, the network including a web page containing the first content.

14. The system of claim 12 , wherein the extracted at least one instance of data is a keyword or key phrase, and wherein the instructions, when executed by the at least one processor, further cause the system to:
locate the keyword or key phrase in the first content; and
match the keyword or key phrase with product-identifying metadata in the first content.

15. The system of claim 12 , wherein the at least one identifier is one of a unique character-based identifier or a hyperlink, and wherein the instructions, when executed by the at least one processor, further cause the system to:
index, in the online marketplace, the at least one identifier and the extracted at least one instance of data.

16. The system of claim 12 , wherein the instructions, when executed by the at least one processor, further cause the system to:
present query results on a client device, the query results based on the association between the extracted instance of data and the at least one entry.

Assignees

• A9 Com Inc

Inventors

• Boteanu Adrian
• Morris Michael James
• Snyder Joshua Christian
• Kiezun Adam
• Gupta Gaurav
• Artzi Shay

Classifications

• G06N3/044

  Recurrent networks, e.g. Hopfield networks · CPC title

• G06N3/045

  Combinations of networks · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N5/02

  Knowledge representation; Symbolic representation · CPC title