Ranking relevant attributes of entity in structured knowledge base

• US 9,229,988 B2
• Filed: 01/18/2013
• Issued: 01/05/2016
• Est. Priority Date: 01/18/2013
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Vadrevu, Srinivas, Tu, Ying, Salvetti, Franco
• Primary Examiner(s): Ruiz, Angelica
• Application Number:
  US13/744,416
• Publication Number:
  US 20140207763A1
• Time in Patent Office: 1,082 Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/288 Entity relationship models
• CPC Class Codes: G06F 16/90 Details of database functions independent of the retrieved data types

Abstract

Architecture that addresses at least the problem of ranking the relevant attributes for a given entity within the context of a structured knowledge base (SKB). The architecture utilizes the attribute, entity type statistics, and the taxonomy of the attributes to consistently and efficiently rank attributes for each and every type of entity in the SKB. Using the SKB, intermediate features are computed, including the importance or popularity each entity type for every entity, inverse document frequency (IDF) computation for each attribute on a global basis, IDF computation for entity types, and the popularity of attributes for each entity type. The intermediate features are aggregated to obtain a final feature set, which can be used in combination with human judgments to train a machine learned classifier model to produce and predict a relevance score for a given entity and each of its attributes. The attributes are ranked for each entity using this score.

Claims

1. A system, comprising:

at least one knowledge base of entities, entity types, and entity attributes;

a features component configured to compute a plurality of intermediate features of the entities and attributes for each entity, and aggregate the intermediate features to output a final feature set of features having an entity-attribute tuple corresponding to the respective plurality of computed intermediate features;

a relevance component configured to generate a relevance score for a given entity and associated attributes based on the feature set, wherein the relevance of each attribute of the entity is based on a set of judgments from human judges;

a ranking component configured to rank the attributes of the given entity based on the relevance scores; and

a microprocessor that executes computer-executable instructions associated with at least one of the features component, relevance component, or the ranking component.

2. The system of claim 1, further comprising a summary component that generates entity summaries that when presented include a brief description of the entity and lists key attributes and associated values in an order of importance to represent and depict for what the entity is popularly and commonly known.

3. The system of claim 1, further comprising a machine learned classifier model that is trained using the feature set to output the relevance scores for the given entity and the associated attributes.

4. The system of claim 3, wherein the classifier model is trained to generate a relevance score for a given entity and associated attribute.

5. The system of claim 3, wherein the classifier model is trained using the final feature set and the set of judgments to generate the relevance score for the given entity and associated attribute.

6. The system of claim 1, wherein the features component computes importance of each entity type for each entity as an intermediate feature, and computes popularity of the attributes for each entity type as an intermediate feature.

7. The system of claim 1, wherein the features component computes term frequency compensation for each attribute on a global basis as an intermediate feature, and term frequency compensation for each entity type as an intermediate feature.

8. A method performed by a computer system executing machine-readable instructions, the method comprising acts of:

receiving one or more structured knowledge bases of entities and attributes;

computing a plurality of intermediate features of the entities and attributes for each entity;

aggregating the intermediate features to output a final feature set of features having an entity-attribute tuple corresponding to the respective plurality of computed intermediate features;

generating a relevance score for a given entity and each of its associated attributes based on the feature set;

ranking the attributes of the given entity based on the relevance scores; and

configuring a microprocessor to execute instructions in a memory associated with at least one of the acts of computing, aggregating, generating, or ranking.

9. The method of claim 8, further comprising generating entity summaries that when presented include a brief description of the entity and lists key attributes and associated values in an order of importance to represent and depict for what the entity is popularly and commonly known.

10. The method of claim 8, further comprising computing importance of each entity type for each entity as an intermediate feature.

11. The method of claim 8, further comprising computing popularity of attributes for each entity type as an intermediate feature.

12. The method of claim 8, further comprising performing term frequency compensation for each attribute on a global basis as an intermediate feature.

13. The method of claim 8, further comprising performing term frequency compensation for each entity type as an intermediate feature.

14. The method of claim 8, further comprising inputting a set of judgments about relevance of each attribute to the associated entity.

15. The method of claim 14, further comprising training a machine learned classifier model to generate a relevance score for a given entity and associated attribute.

16. The method of claim 15, further comprising training the classifier model using the final feature set and the set of judgments to generate the relevance score for the given entity and associated attribute.

17. The method of claim 15, further comprising training the classifier model using the final feature set and an aggregation function to generate the relevance score for the given entity and associated attribute.

18. A method performed by a computer system executing machine-readable instructions, the method comprising acts of:

receiving a structured knowledge base of entities and attributes;

computing intermediate features of the entities and attributes, the intermediate features include importance of each entity type for each entity, popularity of attributes for each entity type, term frequency compensation for each attribute on a global basis, and term frequency compensation for each entity type;

aggregating the intermediate features to output a final feature set of features;

training a classifier model using the final feature set of features, the training based on a set of human judgments from human judges;

generating a relevance score for a given entity and associated attributes based on the classifier model;

ranking the attributes of the given entity based on the relevance scores; and

configuring a microprocessor to execute instructions in a memory associated with at least one of the acts of computing, aggregating, training, generating, or ranking.

19. The method of claim 18, further comprising training the classifier model using the final feature set and the set of human judgments to generate the relevance score for the given entity and associated attribute.

20. The method of claim 18, further comprising training the classifier model using the final feature set and an aggregation function to generate the relevance score for the given entity and associated attribute.