Data store organizing data using semantic classification

• US 9,239,872 B2
• Filed: 10/11/2012
• Issued: 01/19/2016
• Est. Priority Date: 10/11/2012
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Nuance Communications, Inc. (Microsoft Corporation)
• Inventors: Casella dos Santos, Mariana, Montyne, Frank
• Primary Examiner(s): Boccio, Vincent F
• Application Number:
  US13/649,988
• Publication Number:
  US 20140108424A1
• Time in Patent Office: 1,195 Days
• Field of Search: 707/713, 707/739, 707/769, 707/771, 707/748
• US Class Current: 1/1
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06F 16/288 Entity relationship models

Abstract

Data stores that store content units and annotations regarding the content units derived through a semantic interpretation of the content units. When annotations are stored in a database, different parts of an annotation may be stored in different tables of the database. For example, one or more tables of the database may store all semantic classifications for the annotations, while one or more other tables may store content of all of the annotations. A user may be permitted to provide natural language queries for searching the database. A natural language query may be semantically interpreted to determine one or more annotations from the query. The semantic interpretation of the query may be performed using the same annotation model used to determine annotations stored in the database. Semantic classifications and format of the annotations for a query may be the same as one or more annotations stored in the database.

Claims

1. A method for searching a database of information, the database of information comprising information related to a plurality of content units, the information related to the plurality of content units comprising a plurality of stored annotations related to the plurality of content units, the plurality of stored annotations having been generated by processing the plurality of content units using a first semantic interpretation engine, the first semantic interpretation engine processing the plurality of content units according to at least one annotation model, the method comprising:

receiving a query;

determining at least one query annotation relating to the query by processing the query using a second semantic interpretation engine, the second semantic interpretation engine processing the query according to the at least one annotation model, wherein the at least one query annotation comprises a first query annotation, wherein the first query annotation comprises a semantic label for the first query annotation and content of the first query annotation; and

searching the database based at least in part on the first query annotation,

wherein the at least one annotation model of the first and second semantic interpretation engines defines at least one annotation type and, for each annotation type of the at least one annotation type, a semantic label for the annotation type and a format of content for the annotation type, wherein the semantic label for an annotation type defined by the at least one annotation model describes a meaning of content of an annotation having the annotation type in context of a content unit from which the content was extracted and that is annotated by the annotation,

wherein each stored annotation of the plurality of stored annotations is of an annotation type of the at least one annotation type and annotates a content unit of the plurality of content units, each stored annotation of the plurality of stored annotations comprising the semantic label for the annotation type of the stored annotation and content that was extracted from the content unit that the stored annotation annotates and is of the format defined by the at least one annotation model for content of the annotation type,

wherein each query annotation of the at least one query annotation is of an annotation type of the at least one annotation type, each query annotation comprising the semantic label for the annotation type of the query annotation and content that was extracted from the query and is of the format defined by the at least one annotation model for content of the annotation type.

2. The method of claim 1, wherein the determining the at least one query annotation using the second semantic interpretation engine comprises determining a plurality of query annotations, the plurality of query annotations comprising one or more query annotations identifying information having a medical significance and one or more query annotations identifying information that does not have a medical significance.

3. The method of claim 2, wherein the searching the database based at least in part on the first query annotation comprises:

selecting, from the plurality of query annotations, the one or more query annotations identifying information having a medical significance, the one or more query annotations identifying information having a medical significance comprising the first query annotation; and

searching the database based at least in part on the one or more query annotations.

4. The method of claim 1, wherein:

the determining at least one query annotation related to the query comprises determining a plurality of query annotations related to the query using the second semantic interpretation engine; and

the method further comprises;

selecting a selected group of one or more query annotations of the plurality of query annotations based at least in part on a ranking of types of query annotations, wherein selecting the selected group comprises selecting one or more query annotations of one or more types that are ranked higher than one or more types of one or more unselected second query annotations, the selected group comprising the first query annotation; and

searching the database based at least in part on the first query annotation comprises searching the database based at least in part on the selected group of the plurality of query annotations.

5. The method of claim 1, wherein:

the first semantic interpretation engine processes the plurality of content units according to at least one first annotation model that is a first copy of the at least one annotation model that defines the at least one annotation type;

the determining the at least one query annotation relating to the query by processing the query using the second semantic interpretation engine comprises using the second semantic interpretation engine to process the query according to at least one second annotation model, the at least one second annotation model being a second copy of the at least one annotation model defining the at least one annotation type, the second copy of the at least one annotation model being a different copy than the first copy.

6. The method of claim 1, wherein:

the at least one annotation type defined by the at least one annotation model comprises a first plurality of annotation types, each of the first plurality of annotation types defining an annotation having content in a format;

the determining the at least one query annotation relating to the query by processing the query using the second semantic interpretation engine that processes the query according to the at least one annotation model comprises using a second semantic interpretation engine that processes the query according to at least one second annotation model that comprises the at least one annotation model, the at least one second annotation model defining a second plurality of annotation types, the second plurality of annotation types comprising the first plurality of annotation types.

7. The method of claim 1, wherein:

the at least one annotation type defined by the at least one annotation model comprises a plurality of annotation types;

the at least one annotation model is adjusted over time to alter the plurality of annotation types defined by the at least one annotation model;

at least a portion of the plurality of content units were processed by the first semantic interpretation engine at a first time according to the at least one annotation model, the at least one annotation model defining, at the first time, a first plurality of annotation types; and

the determining the at least one query annotation relating to the query by processing the query using the second semantic interpretation engine comprises using, at a second time later than the first time, the second semantic interpretation engine that processes the query according to the at least one annotation model, the at least one annotation model defining, at the second time, a second plurality of annotation types, the second plurality of annotation types defined by the at least one annotation model at the second time being compatible with the first plurality of annotation types defined by the at least one annotation model at the first time.

8. The method of claim 7, wherein the second plurality of annotation types defined by the at least one annotation model at the second time comprises the first plurality of annotation types defined by the at least one annotation model at the first time.

9. The method of claim 1, wherein:

the first semantic interpretation engine and the second semantic interpretation engine are a same semantic interpretation engine; and

the determining the at least one query annotation using the second semantic interpretation engine comprises determining the at least one query annotation using the same semantic interpretation engine executing on a same at least one processor as was used to determine the plurality of stored annotations stored in the database.

10. At least one non-transitory computer-readable storage medium having encoded thereon computer-executable instructions that, when executed by at least one computer, cause the at least one computer to carry out a method for searching a database of information, the database of information comprising information related to a plurality of content units, the information related to the plurality of content units comprising a plurality of stored annotations related to the plurality of content units by processing the plurality of content units using a first semantic interpretation engine, the first semantic interpretation engine processing the plurality of content units according to at least one annotation model, the method comprising:

receiving a query;

determining at least one query annotation relating to the query by processing the query using a second semantic interpretation engine, the second semantic interpretation engine processing the query according to the at least one annotation model, wherein the at least one query annotation comprises a first query annotation, wherein the first query annotation comprises a semantic label for the first query annotation and content of the first query annotation; and

searching the database based at least in part on the first query annotation,

wherein the at least one annotation model of the first and second semantic interpretation engines defines at least one annotation type and, for each annotation type of the at least one annotation type, a semantic label for the annotation type and a format of content for the annotation type, wherein the semantic label for an annotation type defined by the at least one annotation model describes a meaning of content of an annotation having the annotation type in context of a content unit from which the content was extracted and that is annotated by the annotation,

wherein each stored annotation of the plurality of stored annotations is of an annotation type of the at least one annotation type and annotates a content unit of the plurality of content units, each stored annotation of the plurality of stored annotations comprising the semantic label for the annotation type of the stored annotation and content that was extracted from the content unit that the stored annotation annotates and is of the format defined by the at least one annotation model for content of the annotation type,

wherein each query annotation of the at least one query annotation is of an annotation type of the at least one annotation type, each query annotation comprising the semantic label for the annotation type of the query annotation and content that was extracted from the query and is of the format defined by the at least one annotation model for content of the annotation type.

11. The at least one non-transitory computer-readable storage medium of claim 10, wherein the determining the at least one query annotation using the second semantic interpretation engine comprises determining a plurality of query annotations, the plurality of query annotations comprising one or more query annotations identifying information having a medical significance and one or more query annotations identifying information that does not have a medical significance.

12. The at least one non-transitory computer-readable storage medium of claim 11, wherein the searching the database based at least in part on the first query annotation comprises:

selecting, from the plurality of query annotations, the one or more query annotations identifying information having a medical significance, the one or more query annotations identifying information having a medical significance comprising the first query annotation; and

searching the database based at least in part on the first query annotation.

13. The at least one non-transitory computer-readable storage medium of claim 10, wherein:

the first semantic interpretation engine processes the plurality of content units according to at least one first annotation model that is a first copy of the at least one annotation model that defines the at least one annotation type;

the determining the at least one query annotation relating to the query by processing the query using the second semantic interpretation engine comprises using the second semantic interpretation engine to process the query according to at least one second annotation model, the at least one second annotation model being a second copy of the at least one annotation model defining the at least one annotation type, the second copy of the at least one annotation model being a different copy than the first copy.

14. The at least one non-transitory computer-readable storage medium of claim 10, wherein:

the at least one annotation type defined by the at least one annotation model comprises a first plurality of annotation types, each of the first plurality of annotation types defining an annotation having content in a format;

the determining the at least one query annotation relating to the query by processing the query using the second semantic interpretation engine that processes the query according to the at least one annotation model comprises using a second semantic interpretation engine that processes the query according to at least one second annotation model that comprises the at least one annotation model, the at least one second annotation model defining a second plurality of annotation types, the second plurality of annotation types comprising the first plurality of annotation types.

15. The at least one non-transitory computer-readable storage medium of claim 10, wherein:

the at least one annotation type defined by the at least one annotation model comprises a plurality of annotation types;

the at least one annotation model is adjusted over time to alter the plurality of annotation types defined by the at least one annotation model;

at least a portion of the plurality of content units were processed by the first semantic interpretation engine at a first time according to the at least one annotation model, the at least one annotation model defining, at the first time, a first plurality of annotation types; and

the determining the at least one query annotation relating to the query by processing the query using the second semantic interpretation engine comprises using, at a second time later than the first time, the second semantic interpretation engine that processes the query according to the at least one annotation model, the at least one annotation model defining, at the second time, a second plurality of annotation types, the second plurality of annotation types defined by the at least one annotation model at the second time being compatible with the first plurality of annotation types defined by the at least one annotation model at the first time.

16. An apparatus comprising:

at least one processor; and

at least one non-transitory computer-readable storage medium having encoded thereon executable instructions that, when executed by the at least one processor, cause the at least one processor to carry out a method for searching a database of information, the database of information comprising information related to a plurality of content units, the information related to the plurality of content units comprising a plurality of stored annotations related to the plurality of content units by processing the plurality of content units using a first semantic interpretation engine, the first semantic interpretation engine processing the plurality of content units according to at least one annotation model, the method comprising;

receiving a query;

determining at least one query annotation relating to the query by processing the query using a second semantic interpretation engine, the second semantic interpretation engine processing the query according to the at least one annotation model, wherein the at least one query annotation comprises a first query annotation, wherein the first query annotation comprises a semantic label for the first query annotation and content of the first query annotation; and

searching the database based at least in part on the first query annotation

wherein the at least one annotation model of the first and second semantic interpretation engines defines at least one annotation type and, for each annotation type of the at least one annotation type, a semantic label for the annotation type and a format of content for the annotation type, wherein the semantic label for an annotation type defined by the at least one annotation model describes a meaning of content of an annotation having the annotation type in context of a content unit from which the content was extracted and that is annotated by the annotation,

wherein each stored annotation of the plurality of stored annotations is of an annotation type of the at least one annotation type and annotates a content unit of the plurality of content units, each stored annotation of the plurality of stored annotations comprising the semantic label for the annotation type of the stored annotation and content that was extracted from the content unit that the stored annotation annotates and is of the format defined by the at least one annotation model for content of the annotation type,

wherein each query annotation of the at least one query annotation is of an annotation type of the at least one annotation type, each query annotation comprising the semantic label for the annotation type of the query annotation and content that was extracted from the query and is of the format defined by the at least one annotation model for content of the annotation type.

17. The apparatus of claim 16, wherein the determining the at least one query annotation using the second semantic interpretation engine comprises determining a plurality of query annotations, the plurality of query annotations comprising one or more query annotations identifying information having a medical significance and one or more query annotations identifying information that does not have a medical significance.

18. The apparatus of claim 17, wherein the searching the database based at least in part on the first query annotation comprises:

selecting, from the plurality of query annotations, the one or more query annotations identifying information having a medical significance, the one or more query annotations identifying information having a medical significance comprising the first query annotation; and

searching the database based at least in part on the first query annotation.

19. The apparatus of claim 16, wherein:

the first semantic interpretation engine processes the plurality of content units according to at least one first annotation model that is a first copy of the at least one annotation model that defines the at least one annotation type;

the determining the at least one query annotation relating to the query by processing the query using the second semantic interpretation engine comprises using the second semantic interpretation engine to process the query according to at least one second annotation model, the at least one second annotation model being a second copy of the at least one annotation model defining the at least one annotation type, the second copy of the at least one annotation model being a different copy than the first copy.

20. The apparatus of claim 16, wherein:

the at least one annotation type defined by the at least one annotation model comprises a plurality of annotation types;

the at least one annotation model is adjusted over time to alter the plurality of annotation types defined by the at least one annotation model;

at least a portion of the plurality of content units were processed by the first semantic interpretation engine at a first time according to the at least one annotation model, the at least one annotation model defining, at the first time, a first plurality of annotation types; and

the determining the at least one query annotation relating to the query by processing the query using the second semantic interpretation engine comprises using, at a second time later than the first time, the second semantic interpretation engine that processes the query according to the at least one annotation model, the at least one annotation model defining, at the second time, a second plurality of annotation types, the second plurality of annotation types defined by the at least one annotation model at the second time being compatible with the first plurality of annotation types defined by the at least one annotation model at the first time.