Search ranking diversity based on aspect affinity

Patent metadata

Field	Value
Publication number	US-9489461-B2
Application number	US-201414195283-A
Country	US
Kind code	B2
Filing date	Mar 3, 2014
Priority date	Mar 3, 2014
Publication date	Nov 8, 2016
Grant date	Nov 8, 2016

Abstract

Collaborative filtering along the aspect dimension and use of co-occurring aspects to qualify products that will be shown on a search page in response to unspecific user searches, such that diversity in the aspect value will add to the diversity of the search results and the collaborative nature of the aspects will induce diversity, is disclosed. A method and system for Search Ranking Diversity Based on Aspect Affinity includes collecting user search queries, parsing the collected user search queries for aspect phrases, identifying aspect metadata for the aspect phrases, creating a ranked index list of aspects from the aspect metadata, assigning weights to the ranked index list of aspects, receiving a user query and associating the received query with the ranked index list of aspects to produce ranked search results, and re-ranking the search results using diversified aspects from the ranked index list of aspects to produce user search results.

First claim

What is claimed is:

1. A method comprising:
collecting user search queries;
parsing, using one or more hardware processors, the collected user search queries to identify one or more aspect phrases for a product, the aspect phrases qualifying the product;
identifying aspect metadata for the aspect phrases, the aspect metadata indicating a query phrase type for one or more of the aspect phrases;
creating, from the aspect metadata, a list of query phrase types associated with a user query;
assigning weights to the list of query phrase types according to usage of the aspect phrases associated with the query phrase types;
receiving the user query and associating the user query with the list of query phrase types to produce search results; and
ranking the search results using the weights for the list of query phrase types associated with the user query to produce user search results.

2. The method of claim 1 , wherein the ranked index list of aspects is created on a per cluster basis.

3. The method of claim 1 , wherein the ranked index list of aspects is created on a per user basis.

4. The method of claim 1 , wherein the ranked index list of aspects is created on a per user per category basis.

5. The method of claim 1 , wherein the ranked index list of aspects is created on a per user per query basis.

6. The method of claim 1 , wherein re-ranking the search results comprises using a diversified list of values for the most highly correlated k identified important aspects, k being an integer.

7. The method of claim 1 , wherein a user does not explicitly enter a query related to the user search results into a recommender or marketing system.

8. A machine-readable hardware storage medium comprising instructions that, when executed by one or more processors of a machine, cause the machine to perform operations comprising:
collecting a plurality of user search queries for a product;
parsing the collected user search queries to determine one or more aspect phrases for the product, the aspect phrases qualifying the product;
identifying aspect metadata for the aspect phrases, the aspect metadata indicating one or more query phrase types of the aspect phrases;
creating, from the aspect metadata, a list of aspects for a user query;
assigning weights to the list of aspects according to usage of the aspects in the user search queries;
receiving the user query and producing ranked search results using the list of aspects; and
re-ranking the ranked search results using diversified aspects based on the query phrase type(s) from the ranked index list of aspects to produce user search results.

9. The machine-readable hardware storage medium of claim 8 , wherein the ranked index list of aspects is created on a per cluster basis.

10. The machine-readable hardware storage medium of claim 8 , wherein the ranked index list of aspects is created on a per user basis.

11. The machine-readable hardware storage medium of claim 8 , wherein the ranked index list of aspects is created on a per user per category basis.

12. The machine-readable hardware storage medium of claim 8 , wherein the ranked index list of aspects is created on a per user per query basis.

13. The machine-readable hardware storage medium of claim 8 , wherein re-ranking the search results comprises using a diversified list of values for the most highly correlated k identified important aspects, k being an integer.

14. The machine-readable hardware storage medium of claim 8 , wherein a user does not explicitly enter a query related to the user search results into a recommender or marketing system.

15. A system for enabling navigation of a plurality of data items corresponding to a query comprising:
a communication module for receiving a user query for a product;
a search index engine for collecting user search queries for the product; and
an aspect extractor hardware module for parsing collected user search queries to identify one or more aspect phrases for the product, identifying aspect metadata for the aspect phrases, the aspect metadata indicating a query phrase type for one or more of the aspect phrases, creating, from the aspect metadata, a ranked index list of aspects for the user query, the ranked index list of aspects including the user query and one or more query phrase type(s), assigning weights to the ranked index list of aspects according to usage of the aspect phrases associated with the query phrase types, associating the received user query with the ranked index list of aspects to produce ranked search results, and re-ranking the ranked search results using diversified aspects from the ranked index list of aspects to produce user search results.

16. The system of claim 15 , wherein the ranked index list of aspects is created on a per cluster basis.

17. The system of claim 15 , wherein the ranked index list of aspects is created on a per user basis.

18. The system of claim 15 , wherein the ranked index list of aspects is created on a per user per category basis.

19. The system of claim 15 , wherein the ranked index list of aspects is created on a per user per query basis.

20. The system of claim 15 , wherein a user does not explicitly enter a query related to the user search results into a recommender or marketing system.

Assignees

• Ebay Inc

Inventors

• Sundaresan Neelakantan

Classifications

• G06F16/9538

  Presentation of query results · CPC title

• G06F16/24578

  using ranking · CPC title

• G06F16/285

  Clustering or classification · CPC title

• G06F16/951 Primary

  Indexing; Web crawling techniques · CPC title

• G06F16/9536 Primary

  Search customisation based on social or collaborative filtering · CPC title