Granted Patent B2

US 12,260,343 · App. 14/797,858 · Granted Mar 25, 2025

Inventors: Nadiya Yampolska (Kitchener, CA); Mathew Whitney Wilson (Kitchener, CA); Andrew Russell (Duncan, CA); Ihab Francis Ilyas (Waterloo, CA)

Assignee: Primal Fusion Inc.

G06N5/022G06F16/3334G06F16/35G06F16/367G06F16/38G06F16/9024G06F16/955G06F40/205G06Q30/02

Abstract

Claims (91)

1. A method for using knowledge representation construction to augment user context information in a user request, the method comprising:

obtaining user context information associated with a user request, wherein the user context information comprises a first portion and a second portion different from the first portion;

identifying a first group of concepts in a knowledge representation (KR), wherein each concept in the first group of concepts covers a meaning of the user context information, wherein each concept in the first group of concepts is associated with a different meaning of the first portion, wherein each concept in the first group of concepts has a label associated therewith that is substantially the same as each label associated with the other concepts in the first group of concepts;

identifying a second group of concepts in the KR, wherein each concept in the second group of concepts is a semantic context for a respective concept in the first group of concepts;

wherein the identifying is performed at least in part by using at least one processor, and said identifying is based on a measure of dominance and a measure of semantic coherence exceeding a predetermined threshold, and

constructing a context intersection graph using the first group of concepts and the second group of concepts as nodes, and a measure of semantic coherence between the first group of concepts and the second group of concepts as edges, wherein calculating the measure of semantic coherence comprises:

identifying a first semantic context of the first group of concepts in the KR, wherein the first semantic context is defined as a number of concepts in the first group of concepts which are within a predetermined distance of said first concept in said KR; and

identifying a second semantic context of the second group of concepts in the KR, wherein the second semantic context is defined as a number of concepts in the second group of concepts which are within a predetermined distance of said second concept in said KR;

ranking concepts in the context intersection graph based on the measure of semantic coherence of concepts, wherein ranking includes:

calculating a score for each concept in the first group of concepts at least in part by using the context intersection graph; and

identifying the first concept as a concept in the first group of concepts having the highest score; and

augmenting the user request using the user context information and the ranked concepts.

2. The method of claim 1 , further comprising:

calculating the measure of dominance of the first concept using a graph of the KR.

3. The method of claim 1 , wherein the knowledge representation is a semantic network represented by a data structure embodying a directed graph comprising a plurality of nodes and a plurality of edges, wherein each node is associated with a concept in the knowledge representation and an edge between two nodes represents a relationship between concepts associated with the two nodes.

4. The method of claim 1 , wherein the user context information comprises at least one of a search query provided by the user, demographic information about the user, information from the user's browsing history, information typed by the user, and/or information highlighted by the user.

5. A system for using knowledge representation construction to augment user context information in a user request, the system comprising:

at least one processor configured to perform:

obtaining user context information associated with a user request, wherein the user context information comprises a first portion and a second portion different from the first portion;

identifying a first group of concepts in a knowledge representation (KR), wherein each concept in the first group of concepts covers a meaning of the user context information, wherein each concept in the first group of concepts is associated with a different meaning of the first portion, wherein each concept in the first group of concepts has a label associated therewith that is substantially the same as each label associated with the other concepts in the first group of concepts;

identifying a second group of concepts in the KR, wherein each concept in the second group of concepts is a semantic context for a respective concept in the first group of concepts;

wherein the identifying is performed at least in part by using at least one processor, and said identifying is based on a measure of dominance and a measure of semantic coherence exceeding a predetermined threshold, and

constructing a context intersection graph using the first group of concepts and the second group of concepts as nodes, and a measure of semantic coherence between the first group of concepts and the second group of concepts as edges, wherein calculating the measure of semantic coherence comprises:

identifying a first semantic context of the first group of concepts in the KR, wherein the first semantic context is defined as a number of concepts in the first group of concepts which are within a predetermined distance of said first concept in said KR; and

identifying a second semantic context of the second group of concepts in the KR, wherein the second semantic context is defined as a number of concepts in the second group of concepts which are within a predetermined distance of said second concept in said KR;

ranking concepts in the context intersection graph based on the measure of semantic coherence of concepts, wherein ranking includes:

calculating a score for each concept in the first group of concepts at least in part by using the context intersection graph; and

identifying the first concept as a concept in the first group of concepts having the highest score; and

augmenting the user request using the user context information and the ranked concepts.

6. At least one non-transitory computer-readable storage medium storing processor executable instructions that, when executed by at least one processor, cause the at least one processor to perform a method comprising:

obtaining user context information associated with a user request, wherein the user context information comprises a first portion and a second portion different from the first portion;

identifying a first group of concepts in a knowledge representation (KR), wherein each concept in the first group of concepts covers a meaning of the user context information, wherein each concept in the first group of concepts is associated with a different meaning of the first portion, wherein each concept in the first group of concepts has a label associated therewith that is substantially the same as each label associated with the other concepts in the first group of concepts;

identifying a second group of concepts in the KR, wherein each concept in the second group of concepts is a semantic context for a respective concept in the first group of concepts;

wherein the identifying is performed at least in part by using at least one processor, and said identifying is based on a measure of dominance and a measure of semantic coherence exceeding a predetermined threshold, and

constructing a context intersection graph using the first group of concepts and the second group of concepts as nodes, and a measure of semantic coherence between the first group of concepts and the second group of concepts as edges, wherein calculating the measure of semantic coherence comprises:

identifying a first semantic context of the first group of concepts in the KR, wherein the first semantic context is defined as a number of concepts in the first group of concepts which are within a predetermined distance of said first concept in said KR; and

identifying a second semantic context of the second group of concepts in the KR, wherein the second semantic context is defined as a number of concepts in the second group of concepts which are within a predetermined distance of said second concept in said KR;

ranking concepts in the context intersection graph based on the measure of semantic coherence of concepts, wherein ranking includes:

calculating a score for each concept in the first group of concepts at least in part by using the context intersection graph; and

identifying the first concept as a concept in the first group of concepts having the highest score; and

augmenting the user request using the user context information and the ranked concepts.

7. A method comprising:

obtaining user context information associated with a user request, wherein the user context information comprises a first portion and a second portion different from the first portion;

disambiguating between a first group of concepts in a knowledge representation (KR) associated with a first meaning of the first portion and a second group of concepts in the KR,

wherein each concept in the first group of concepts covers a meaning of the user context information and each concept in the second group of concepts is a semantic context for a respective concept in the first group of concepts;

wherein the disambiguating is performed at least in part by using at least one processor, a measure of dominance and a measure of semantic coherence exceeding a predetermined threshold,

constructing a context intersection graph using the first group of concepts and the second group of concepts as nodes, and a measure of semantic coherence between the first group of concepts and the second group of concepts as edges concept, wherein calculating the measure of semantic coherence comprises:

identifying a first semantic context of the first group of concepts in the KR, wherein the first semantic context is defined as a number of concepts in the first group of concepts which are within a predetermined distance of said first concept in said KR; and

identifying a second semantic context of the second group of concepts in the KR, wherein the second semantic context is defined as a number of concepts in the second group of concepts which are within a predetermined distance of said second concept in said KR;

ranking concepts in the context intersection graph based on the measure of semantic coherence of concepts, wherein ranking includes:

calculating a score for each concept in the first group of concepts at least in part by using the context intersection graph; and

identifying the first concept as a concept in the first group of concepts having the highest score; and

augmenting the user request using the user context information and the ranked concepts.

8. A system comprising:

at least one processor configured to perform:

obtaining user context information associated with a user request, wherein the user context information comprises a first portion and a second portion different from the first portion;

disambiguating between a first group of concepts in a knowledge representation (KR) associated with a first meaning of the first portion and a second group of concepts in the KR,

wherein each concept in the first group of concepts covers a meaning of the user context information and each concept in the second group of concepts is a semantic context for a respective concept in the first group of concepts;

wherein the disambiguating is performed at least in part by using a measure of dominance and a measure of semantic coherence exceeding a predetermined threshold,

constructing a context intersection graph using the first group of concepts and the second group of concepts as nodes, and a measure of semantic coherence between the first group of concepts and the second group of concepts as edges concept, wherein calculating the measure of semantic coherence comprises:

identifying a first semantic context of the first group of concepts in the KR, wherein the first semantic context is defined as a number of concepts in the first group of concepts which are within a predetermined distance of said first concept in said KR; and

identifying a second semantic context of the second group of concepts in the KR, wherein the second semantic context is defined as a number of concepts in the second group of concepts which are within a predetermined distance of said second concept in said KR;

ranking concepts in the context intersection graph based on the measure of semantic coherence of concepts, wherein ranking includes:

calculating a score for each concept in the first group of concepts at least in part by using the context intersection graph; and

identifying the first concept as a concept in the first group of concepts having the highest score; and

augmenting the user request using the user context information and the ranked concepts.

9. At least one non-transitory computer-readable storage medium storing processor executable instructions that, when executed by at least one processor, cause the at least one processor to perform a method comprising:

obtaining user context information associated with a user request, wherein the user context information comprises a first portion and a second portion different from the first portion;

disambiguating between a first group of concepts in a knowledge representation (KR) associated with a first meaning of the first portion and a second group of concepts in the KR,

wherein each concept in the first group of concepts covers a meaning of the user context information and each concept in the second group of concepts is a semantic context for a respective concept in the first group of concepts;

wherein the disambiguating is performed at least in part by using a measure of dominance and a measure of semantic coherence exceeding a predetermined threshold,

constructing a context intersection graph using the first group of concepts and the second group of concepts as nodes, and a measure of semantic coherence between the first group of concepts and the second group of concepts as edges concept, wherein calculating the measure of semantic coherence comprises:

identifying a first semantic context of the first group of concepts in the KR, wherein the first semantic context is defined as a number of concepts in the first group of concepts which are within a predetermined distance of said first concept in said KR; and

identifying a second semantic context of the second group of concepts in the KR, wherein the second semantic context is defined as a number of concepts in the second group of concepts which are within a predetermined distance of said second concept in said KR;

ranking concepts in the context intersection graph based on the measure of semantic coherence of concepts, wherein ranking includes:

calculating a score for each concept in the first group of concepts at least in part by using the context intersection graph; and

identifying the first concept as a concept in the first group of concepts having the highest score; and

augmenting the user request using the user context information and the ranked concepts.

10. The method of claim 1 , further comprising calculating an amount of overlap between the first semantic context and the second semantic context using Dice's coefficient.

11. The method of claim 1 , further comprising calculating an amount of overlap based on labels of concepts in the first semantic context and labels of concepts in the second semantic context.

12. The method of claim 1 , wherein identifying the first semantic context comprises identifying at least one direct attribute of the first concept in the KR.

13. The method of claim 1 , wherein identifying the first semantic context comprises identifying at least one indirect attribute of the first concept in the KR.

14. The method of claim 1 , wherein identifying the first semantic context comprises identifying at least one ancestor of the first concept in the KR.

15. The method of claim 1 , wherein the first semantic context comprises at least one direct attribute of the first concept, at least one ancestor of the first concept, and at least one indirect attribute of the first concept.

16. The method of claim 1 , wherein the first semantic context comprises concepts within a predetermined distance of the first concept on the graph of the KR.

17. The method of claim 1 , wherein the knowledge representation is a semantic network represented by a data structure embodying a directed graph comprising a plurality of nodes and a plurality of edges, wherein each node is associated with a concept in the knowledge representation and an edge between two nodes represents a relationship between concepts associated with the two nodes.

18. The method of claim 7 , wherein the user context information comprises at least one of a search query provided by the user, demographic information about the user, information from the user's browsing history, information typed by the user, and/or information highlighted by the user.

19. The system of claim 5 , wherein the processor is further configured to perform:

calculating the measure of dominance of the first concept using a graph of the KR.

20. The system of claim 5 , wherein the knowledge representation is a semantic network represented by a data structure embodying a directed graph comprising a plurality of nodes and a plurality of edges, wherein each node is associated with a concept in the knowledge representation and an edge between two nodes represents a relationship between concepts associated with the two nodes.

21. The system of claim 5 , wherein the user context information comprises at least one of a search query provided by the user, demographic information about the user, information from the user's browsing history, information typed by the user, and/or information highlighted by the user.

Assignments (3)

Continuity (6)

References Cited (82)