Dynamic faceting for personalized search and discovery

Patent metadata

Field	Value
Publication number	US-10430465-B2
Application number	US-201715398224-A
Country	US
Kind code	B2
Filing date	Jan 4, 2017
Priority date	Jan 4, 2017
Publication date	Oct 1, 2019
Grant date	Oct 1, 2019

Abstract

Methods, computer program products, and systems are presented. The methods include, for instance: determining user clusters and navigation-type clusters based on multiple information requests, and training facets and corresponding usefulness factor of the facets from the multiple information requests by machine learning. When a user submits a query, the user and the query is respectively mapped with one of the user clusters and the navigation-type clusters, and the query is customized based on the associated pair of clusters. Results of the query are obtained, ranked by usefulness of the facets as determined according to the pair of clusters, and presented to the user.

First claim

What is claimed is:

1. A computer implemented method for personalizing a query by use of dynamic faceting, comprising:
identifying, by one or more processor of a computer, at least one user cluster based on respective intent from multiple requests for information including searches and discoveries, wherein each intent corresponds to a specific characterization of users in respective user cluster;
specifying at least one navigation-type cluster based on respective navigation pattern from the multiple requests for information;
training, by machine learning, at least one facet and usefulness factors respectively associated with the at least one facet, as determined for each pair of a user cluster and a navigation-type cluster, wherein the user cluster is one of the at least one user cluster and the navigation-type cluster is one of the at least one navigation-type cluster, and presenting the at least one facet to at least one user belonging to the user cluster;
customizing a query received from a user by use of one or more selected facet and corresponding associated usefulness factor, the one or more selected facet corresponding to a selected pair from the at least one user cluster and the at least one navigation-type cluster such that the user receives a response most appropriate for an intent of the query; and
producing results responsive to the query, wherein facets amongst the at least one facet in the results are scored according to the usefulness factor respectively corresponding to the one or more selected facet.

2. The computer implemented method of claim 1 , the customizing comprising:
mapping the user to a first user cluster from the at least one user cluster, wherein the first user cluster represents attributes most similar to the user amongst the at least one user cluster;
associating the query to a first navigation-type cluster from the at least one navigation-type cluster, wherein the first navigation-type cluster represents attributes most similar to the intent of the query amongst the at least one navigation-type cluster; and
displaying, to the user, the one or more facet corresponding to the selected pair of the first user cluster and the first navigation-type cluster, wherein the one or more selected facet is ranked according to the one or more usefulness factor respectively corresponding to the one or more selected facet.

3. The computer implemented method of claim 2 , further comprising:
receiving, from the user, a feedback on the one or more selected facet corresponding to the selected pair of the first user cluster and the first navigation-type cluster such that the respective usefulness factors of the one or more selected facet may be adjusted based on the feedback.

4. The computer implemented method of claim 1 , the training comprising:
reviewing successful results including search histories, navigation histories, search terms and search topics, by machine learning, for the one or more selected facet and the usefulness factor respectively associated with the one or more selected facet, wherein the successful results are determined by various user responses including user comments and long-click on the results.

5. The computer implemented method of claim 1 , further comprising:
deriving candidate facets to be the one or more selected facet for the selected pair from the at least one user cluster and the at least one navigation-type cluster by machine learning; and
rebuilding one or more search index partially or fully with the candidate facets such that another query similar to the query received from another user similar to the user can be more efficiently processed with the rebuilt one or more search index, wherein the rebuilt search index includes the candidate facets as well as the one or more selected facet.

6. The computer implemented method of claim 1 ,
wherein the respective intent for determining the at least one user cluster is selected from the group consisting of: personality traits, demographic data, ages, geographic locations, gender, group memberships, occupations, psychographic, chronotype, and/or socioeconomic categories of the users, device types, transaction histories, and combinations thereof.

7. The computer implemented method of claim 1 ,
wherein the respective navigation pattern for determining the at least one navigation-type cluster is selected from the group consisting of: navigation activities and search queries from various data sources including one of cookies, browsing histories, purchase histories, online transaction histories, sequences of searches, current contexts of searches as derived from previous activities and interactions of the users, social media footprint data of the users, envelope information of the searches including time of the searches and locations of the users, clickstreams representing click activities on the screen by the users, survey data, call center data generated in association with the activities of the users, and combinations thereof.

8. A computer program product comprising:
a computer readable storage medium readable by one or more processor and storing instructions for execution by the one or more processor for performing a method for personalizing a query by use of dynamic faceting, comprising:
identifying, by the one or more processor, at least one user cluster based on respective intent from multiple requests for information including searches and discoveries, wherein each intent corresponds to a specific characterization of users in respective user cluster;
specifying at least one navigation-type cluster based on respective navigation pattern from the multiple requests for information;
training, by machine learning, at least one facet and usefulness factors respectively associated with the at least one facet, as determined for each pair of a user cluster and a navigation-type cluster, wherein the user cluster is one of the at least one user cluster and the navigation-type cluster is one of the at least one navigation-type cluster, and presenting the at least one facet to at least one user belonging to the user cluster;
customizing a query received from a user by use of one or more selected facet and corresponding associated usefulness factor, the one or more selected facet corresponding to a selected pair from the at least one user cluster and the at least one navigation-type cluster such that the user receives a response most appropriate for an intent of the query; and
producing results responsive to the query, wherein facet amongst the at least one facet in the results are scored according to the usefulness factor respectively corresponding to the one or more selected facet.

9. The computer program product of claim 8 , the customizing comprising:
mapping the user to a first user cluster from the at least one user cluster, wherein the first user cluster represents attributes most similar to the user amongst the at least one user cluster;
associating the query to a first navigation-type cluster from the at least one navigation-type cluster, wherein the first navigation-type cluster represents attributes most similar to the intent of the query amongst the at least one navigation-type cluster; and
displaying, to the user, the one or more facet corresponding to the selected pair of the first user cluster and the first navigation-type cluster, wherein the one or more selected facet is ranked according to the one or more usefulness factor respectively corresponding to the one or more selected facet.

10. The computer program product of claim 9 , further comprising:
receiving, from the user, a feedback on the one or more selected facet corresponding to the selected pair of the first user cluster and the first navigat

Assignees

• IBM

Inventors

• Freed Andrew R
• Herman Norbert
• Ray Shubhadip
• Sanyal Avik

Classifications

• G06F16/9032 Primary

  Query formulation · CPC title

• G06N20/00 Primary

  Machine learning · CPC title

• G06F18/23

  Clustering techniques · CPC title

• G06F16/90335

  Query processing · CPC title