EXPLOITING THE SEMANTIC WEB FOR UNSUPERVISED NATURAL LANGUAGE SEMANTIC PARSING

• US 20140236575A1
• Filed: 02/21/2013
• Published: 08/21/2014
• Est. Priority Date: 02/21/2013
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Tur, Gokhan, Hakkani-Tur, Dilek, Heck, Larry, Wang, Ye-Yi, Jeong, Minwoo
• Granted Patent: US 10,235,358 B2
• Time in Patent Office: Days
• US Class Current: 704/9
• CPC Class Codes: G06F 40/30 Semantic analysis

Abstract

Structured web pages are accessed and parsed to obtain implicit annotation for natural language understanding tasks. Search queries that hit these structured web pages are automatically mined for information that is used to semantically annotate the queries. The automatically annotated queries may be used for automatically building statistical unsupervised slot filling models without using a semantic annotation guideline. For example, tags that are located on a structured web page that are associated with the search query may be used to annotate the query. The mined search queries may be filtered to create a set of queries that is in a form of a natural language query and/or remove queries that are difficult to parse. A natural language model may be trained using the resulting mined queries. Some queries may be set aside for testing and the model may be adapted using in-domain sentences that are not annotated. The models may be tested using these implicitly annotated natural-language-like queries in an unsupervised fashion.

Claims

1. A method for natural language semantic parsing, comprising:

accessing structured content;

parsing the structured content;

automatically mining search queries that hit to the structured content; and

automatically annotating queries obtained from automatically mining the search queries.

2. The method of claim 1, wherein the structured content are structured web pages.

3. The method of claim 1, further comprising building an unsupervised slot filling model using the annotated queries.

4. The method of claim 3, further comprising adapting the unsupervised slot filling model using in-domain unannotated sentences.

5. The method of claim 4, further comprising testing a performance of the model using natural-language like queries determined from the results obtained from automatically mining the search queries.

6. The method of claim 1, wherein the structured content is defined by a triple that consists of two entities linked by a relation.

7. The method of claim 1, further comprising filtering the queries obtained from automatically mining the search queries to determine the queries that are in a form of a natural language query.

8. The method of claim 1, further comprising semantically annotating the queries using tags in a structured web page that the query hits.

9. A computer-readable medium storing computer-executable instructions for natural language semantic parsing, comprising:

accessing structured content;

parsing the structured content;

automatically mining search queries that hit to the structured content;

automatically annotating queries obtained from automatically mining the search queries; and

creating an understanding model including slots using the annotated queries.

10. The computer-readable medium of claim 9, wherein the structured content are structured web pages.

11. The computer-readable medium of claim 10, wherein the understanding model is created in an unsupervised manner.

12. The computer-readable medium of claim 10, further comprising testing a performance of the model using natural-language like queries obtained from automatically mining the search queries.

13. The computer-readable medium of claim 10, wherein the structured content is defined by a triple that consists of two entities linked by a relation.

14. The computer-readable medium of claim 10, further comprising filtering the queries obtained from automatically mining the search queries to determine the queries that are in a form of a natural language query.

15. The computer-readable medium of claim 10, further comprising semantically annotating the queries using tags in a structured web page that the query hits.

16. A system for natural language semantic parsing, comprising:

a processor and memory;

an operating environment executing using the processor; and

a knowledge manager that is configured to perform actions comprising;

accessing structured web pages;

parsing the structured web pages;

automatically mining search queries that hit to the structured web pages;

automatically annotating queries obtained from automatically mining the search queries; and

creating an understanding model including slots using the annotated queries.

17. The system of claim 16, wherein the understanding model is created in an unsupervised manner.

18. The system of claim 16, further comprising testing a performance of the model using natural-language like queries obtained from automatically mining the search queries.

19. The system of claim 16, wherein the structured content is defined by a triple that consists of two entities linked by a relation.

20. The system of claim 16, further comprising filtering the queries obtained from automatically mining the search queries to determine the queries that are in a form of a natural language query.