ENTITY AND ATTRIBUTE RESOLUTION IN CONVERSATIONAL APPLICATIONS

• US 20150286747A1
• Filed: 04/02/2014
• Published: 10/08/2015
• Est. Priority Date: 04/02/2014
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Anastasakos, Tasos, Rochette, Alexandre, Sarikaya, Ruhi
• Granted Patent: US 9,892,208 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/2465 Query processing support for facilitating data mining operations in structured databases
• CPC Class Codes: G06F 16/313 Selection or weighting of terms for indexing
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 16/86 Mapping to a database
• CPC Class Codes: G06F 16/9017 using directory or table look-up use of a directory or look-up table in file systems G06F16/13
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/9538 Presentation of query results
• CPC Class Codes: G06F 40/20 Natural language analysis semantic analysis of natural language G06F40/30
• CPC Class Codes: G06F 40/40 Processing or translation of natural language natural language analysis G06F40/20; semantic analysis G06F40/30

Abstract

User input expressed as text may be analyzed for determining a type of response, such as an application response, and/or determining a type of task that is requested by the user input. Entity representations may be identified, classified and/or or tagged based on a type of response, type of task and/or a set of entity types. A surface form of an entity, ambiguous entity representation and/or other type of expression within the user input may be resolved, normalized and/or mapped to a normalized value. Normalizing entities and/or entity attributes may involve using a set of normalization rules, a lookup table, one or more machined learned methods, and/or an entity normalization index that associates entities with alternate surface forms derived from web corpora. The normalized value may be used to construct a request to a structured knowledge source and/or an application.

Claims

1. A computer-implemented method comprising:

receiving user input expressed as text at a computer system;

analyzing text contained in the user input to determine a type of application response that is requested by the user input;

identifying one or more entity representations within the user input based on a set of entity types utilized by an application that is configured to provide the type of application response;

detecting an entity representation that describes an entity with a surface form of the entity;

mapping the surface form of the entity to a normalized value; and

constructing a request to the application using the normalized value.

2. The computer-implemented method of claim 1, wherein analyzing the text contained in the user input includes one or more of:

detecting a domain of the user input, and

detecting an intent of the user input.

3. The computer-implemented method of claim 1, wherein the set of entity types corresponds to a set of domain-dependent entity attributes.

4. The computer-implemented method of claim 1, further comprising:

tagging entity representations contained in the user input with slot types corresponding to the set of entity types.

5. The computer-implemented method of claim 1, wherein:

mapping the surface form to the normalized value is performed deterministically using a set of normalization rules to normalize one of a temporal expression, a geographic expression, or a numerical expression to a canonical format.

6. The computer-implemented method of claim 1, wherein mapping the surface form to the normalized value is performed using a lookup table to normalize a variant of the entity to a canonical name.

7. The computer-implemented method of claim 1, wherein mapping the surface form to the normalized value is performed using one or more machined learned methods to map a natural language expression to a normalized condition.

8. The computer-implemented method of claim 1, wherein mapping the surface form to the normalized value is performed using an entity normalization index that associates entities with alternate surface forms derived from web corpora.

9. The computer-implemented method of claim 8, further comprising:

extracting a set of entities from an entity knowledge source;

extracting related links that are relevant to the set of entities from a knowledge graph;

data mining one or more web search logs using the related links to discover related queries; and

building the entity normalization index by associating the related queries with the set of entities as alternate surface forms of the entities.

10. The computer-implemented method of claim 1, wherein the normalized value is used to construct a request that is configured for interacting with a structured knowledge source.

11. The computer-implemented method of claim 1, wherein the normalized value is used to construct a request that causes a task-based application to perform a particular task.

12. A computer-readable storage medium storing computer-readable instructions that, when executed by a processor of a computing device, cause the computing device to perform a method comprising:

analyzing user input expressed as text to determine a type of response that is requested by the user input;

tagging one or more entity representations within the user input based on a set of entity types associated with the type of response;

detecting a surface form of an entity within a tagged entity representation;

normalizing the surface form to a canonical form of the entity; and

constructing a request using the canonical form of the entity.

13. The computer-readable storage medium of claim 12, wherein normalizing the surface form comprises using one or more of:

a set of normalization rules, a lookup table, a machine learned method, or an index that associates entities with alternate surface forms derived from web corpora.

14. The computer-readable storage medium of claim 12, wherein the surface form is one or more of:

a temporal expression, a geographic expression, a numerical expression, or a natural language expression.

15. The computer-readable storage medium of claim 12, wherein the request is configured for one or more of:

interacting with a structured knowledge source using canonical values supported by the structured knowledge source, causing an application to perform a particular task, or querying a remote resource for information related to the user input.

16. The computer-readable storage medium of claim 12, wherein the set of entity types corresponds to a set of domain-dependent entity attributes.

17. A computer system comprising:

a processor configure to execute computer-readable instructions; and

memory storing computer-readable instructions configured to;

analyze text contained in user input to determine a type of task that is requested by the user input;

classify sub-phrases contained in the text into one or more entity representations based on the type of task;

detect an ambiguous entity representation;

resolve the ambiguous entity representation to a normalized value; and

construct a request for a particular task using the normalized value.

18. The computer system of claim 17, wherein the memory further stores an application that is configured to perform the particular task.

19. The computer system of claim 17, wherein the memory further stores computer-readable instructions for communicating the request to an application that is configured to perform the particular task.

20. The computer system of claim 17, wherein the particular task is one or more of:

answering a question, playing a media item, or scheduling an event.