Systems and methods for updating a knowledge graph through user input

• US 10,078,651 B2
• Filed: 04/27/2015
• Issued: 09/18/2018
• Est. Priority Date: 04/27/2015
• Status: Active Grant

• Current Assignee: Rovi Guides, Inc. (Adeia Inc.)
• Original Assignee: Rovi Guides, Inc. (Adeia Inc.)
• Inventors: Kelly, Timothy Christensen, Morgan, Patrick Gerard, Peterson, Brian, Klappert, Walter R.
• Primary Examiner(s): Thai, Hanh B
• Application Number:
  US14/697,196
• Publication Number:
  US 20160314158A1
• Time in Patent Office: 1,240 Days
• Field of Search: 707736-737, 707771, 707798, 707804
• CPC Class Codes: G06F 16/217 Database tuning G06F16/2282 takes precedence; database performance monitoring G06F11/3409
• CPC Class Codes: G06F 16/23 Updating
• CPC Class Codes: G06F 16/367 Ontology
• CPC Class Codes: G06F 40/279 Recognition of textual entities

Abstract

Methods and systems are disclosed herein for updating a knowledge graph based on a user confirmation. A media guidance application receives a user communication and isolates a term of the user communication. The media guidance application identifies a candidate component of a knowledge graph associated with the term. The media guidance application requests user input directed to confirming whether the term is associated with the candidate component. In response to receiving the user input, the media guidance application modifies a strength of association between the term and the component.

Claims

1. A method for providing recommendations that take into account users'"'"' casual references to certain media assets during conversational communications by isolating a term in the conversational communications and determining relationships to that term through the use of a knowledge graph organized to store relationships between different pieces of information, comprising:

receiving, via a user interface, a user communication;

analyzing, via processing circuitry, the user communication based on a previously stored template from a database to determine a term of the user communication;

transmitting a query based on the term to a knowledge graph having a plurality of nodes representing different pieces of information and a plurality of edges connecting the plurality of nodes representing relationships between the different pieces of information;

identifying a first node representing a candidate component that is connected to a second node representing the term based on the knowledge graph;

generating for display, via the user interface, a request causing user input directed to confirming whether the term is associated with the candidate component;

in response to receiving the user input, modifying a strength of association represented by a weight allocated to an edge connecting the second node representing the term and the first node representing the candidate component in the knowledge graph; and

generating a content recommendation associated with the candidate component based on the strength of association.

2. The method of claim 1, wherein the user input is further directed to confirming whether the strength of association between the term and the candidate component should be changed.

3. The method of claim 1, wherein modifying the strength of association between the term and the candidate component further comprises:

determining whether the user input confirms or denies that the term is associated with the candidate component;

in response to determining that the user input confirms that the term is associated with the candidate component, increasing the strength of association between the term and the candidate component; and

in response to determining that the user input denies that the term is associated with the candidate component, decreasing the strength of association between the term and the candidate component.

4. The method of claim 1, wherein modifying the strength of association between the term and the candidate component further comprises:

determining that the term is associated with a first node in the knowledge graph;

determining that the candidate component is associated with a second node in the knowledge graph; and

causing a value associated with an edge connecting the first node and the second node to be changed.

5. The method of claim 1, wherein modifying the strength of association between the term and the candidate component further comprises:

generating a new component of the knowledge graph corresponding to the term; and

generating an association between the new component and the candidate component.

6. The method of claim 1, wherein requesting the user input further comprises providing information associated with the candidate component.

7. The method of claim 1, further comprising providing information associated with an additional component of the knowledge graph that has high strength of association between the term and the additional component.

8. The method of claim 1, wherein identifying the candidate component further comprises cross-referencing the term with a plurality of candidate components of the knowledge graph stored in a database.

9. The method of claim 1, wherein the user input is a first user input, and wherein identifying the candidate component further comprises:

identifying a plurality of candidate components of the knowledge graph;

ranking each of the plurality of candidate components;

identifying a highest ranked candidate component of the plurality of candidate components as the candidate component;

receiving a second user input that denies that the term is associated with the candidate component; and

in response to receiving the second user input, identifying a next highest ranked candidate component of the plurality of candidate components as the candidate component.

10. The method of claim 1, wherein the user communication may be verbal or textual.

11. A system for providing recommendations that take into account users'"'"' casual references to certain media assets during conversational communications by isolating a term in the conversational communications and determining relationships to that term through the use of a knowledge graph organized to store relationships between different pieces of information, comprising:

storage circuitry configured to store a knowledge graph having a plurality of nodes representing different pieces of information and a plurality of edges connecting the plurality of nodes representing relationships between the different pieces of information;

communications circuitry configured to receive a user communication; and

control circuitry configured to;

analyze the user communication based on a previously stored template from a database to determine a term of the user communication;

transmit a query based on the term to a knowledge graph;

identify a first node representing a candidate component that is connected to a second node representing the term based on the knowledge graph;

generating for display, via the user interface, a request causing user input directed to confirming whether the term is associated with the candidate component;

in response to receiving the user input, modify a strength of association represented by a weight allocated to an edge connecting the second node representing the term and the first node representing the candidate component in the knowledge graph; and

generate a content recommendation associated with the candidate component based on the strength of association.

12. The system of claim 11, wherein the user input is further directed to confirming whether the strength of association between the term and the candidate component should be changed.

13. The system of claim 11, wherein the control circuitry is further configured, when modifying the strength of association between the term and the candidate component to:

determine whether the user input confirms or denies that the term is associated with the candidate component;

in response to determining that the user input confirms that the term is associated with the candidate component, increase the strength of association between the term and the candidate component; and

in response to determining that the user input denies that the term is associated with the candidate component, decrease the strength of association between the term and the candidate component.

14. The system of claim 11, wherein the control circuitry is further configured, when modifying the strength of association between the term and the candidate component to:

determine that the term is associated with a first node in the knowledge graph;

determine that the candidate component is associated with a second node in the knowledge graph; and

cause a value associated with an edge connecting the first node and the second node to be changed.

15. The system of claim 11, wherein the control circuitry is further configured, when modifying the strength of association between the term and the candidate component to:

generate a new component of the knowledge graph corresponding to the term; and

generate an association between the new component and the candidate component.

16. The system of claim 11, wherein the control circuitry is further configured, when requesting the user input, to provide information associated with the candidate component.

17. The system of claim 11, wherein the control circuitry is further configured to provide information associated with an additional component of the knowledge graph that has high strength of association between the term and the additional component.

18. The system of claim 11, wherein the control circuitry is further configured, when identifying the candidate component, to cross-reference the term with a plurality of candidate components of the knowledge graph stored in a database.

19. The system of claim 11, wherein the user input is a first user input, and wherein the control circuitry is further configured, when identifying the candidate, to:

identify a plurality of candidate components of the knowledge graph;

rank each of the plurality of candidate components;

identify a highest ranked candidate component of the plurality of candidate components as the candidate component;

receive a second user input that denies that the term is associated with the candidate component; and

in response to receiving the second user input, identify a next highest ranked candidate component of the plurality of candidate components as the candidate component.

20. The system of claim 11, wherein the user communication may be verbal or textual.