Systems and methods for answering user questions

• US 20070022109A1
• Filed: 05/16/2006
• Published: 01/25/2007
• Est. Priority Date: 07/25/2005
• Status: Active Grant

• Current Assignee: IAC Search & Media Incorporated (Match Group, Inc.)
• Original Assignee: IAC Search & Media Incorporated (Match Group, Inc.)
• Inventors: Imielinski, Tomasz, Mittal, Rohit, Goyal, Jay, Zola, Scot
• Granted Patent: US 8,756,245 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/243 Natural language query formulation
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems
• CPC Class Codes: G06F 16/3344 using natural language analysis

Abstract

Systems and methods for answering user questions using databases annotated with answerable questions are disclosed. The database includes database items organized into columns. Each of the columns may have a question annotated thereto. The question is a parameterized question that is answerable by the database items in the column. A question answering algorithm is used to search the database for an annotated question corresponding to a question that needs to be answered.

Claims

1. A method for natural language searching comprising:

receiving a search query in the form of a natural language question;

normalizing the search query;

searching a database having a plurality of database items and questions tagged to the database associated with the database items for a question that matches the normalized search query; and

providing an answer corresponding to the database item and question that matches the normalized search query.

2. The method of claim 1, wherein the question is a parameterized question.

3. The method of claim 1, wherein the database items are arranged into a plurality of columns and each of the columns has a question tagged to the column.

4. The method of claim 3, wherein the question tagged to the column is a question answerable by the database items in the column.

5. The method of claim 1, wherein the search query and the tagged question are exact matches.

6. The method of claim 1, wherein the tagged question has the answer associated therewith.

7. The method of claim 6, wherein the answer is a parameterized answer, and at least one database item is substituted into the parameterized answer.

8. The method of claim 1, wherein normalizing the search query comprises one or more selected from the group consisting of removing unnecessary words from the search, query, rephrasing the question into an answerable question, identifying and replacing words with synonyms, and modifying pluralities into singularities.

9. The method of claim 1, wherein the search query is matched to the tagged question by:

automatically extracting features from the search query;

identifying a database to be searched from the features; and

identifying a column with a tagged question from the features.

10. The method of claim 1, wherein the search query is matched to the tagged question by:

automatically extracting features from the search query;

mapping at least one of the features to a concept corresponding to the database; and

mapping the question that needs to be answered to a matching tagged question.

11. The method of claim 1, further comprising identifying a database to search.

12. The method of claim 11, wherein identifying a database to search comprises comparing concepts extracted from the search query with concepts related to the database items in the database.

13. A natural language search system comprising:

a search engine to receive a search query in the form of a natural language question;

a database to store a plurality of database items organized into columns and a question annotated to each of the columns, the question being a parameterized question that is answerable by the database items in the column; and

a server to search the database for a question that matches the search query and provide an answer corresponding to the database item and parameterized question to the search engine.

14. The system of claim 13, wherein the answer is a parameterized answer, and the server substitutes database items corresponding to the answer to the search query into the parameterized answer before the answer is provided to the search engine.

15. The system of claim 13, wherein the server is further to normalize the search query.

16. The system of claim 13, wherein the search query and annotated question are an exact match.

17. The system of claim 13, wherein the server is further to identify a database to search.

18. A question answering system comprising:

a database having a plurality of database items organized into columns, each of the columns having a question annotated thereto, the question being a parameterized question that is answerable by the database items in the column;

a question answering interface, in communication with the database, the question answering interface having a question answering algorithm that searches the database for an annotated question corresponding to a question that needs to be answered.

19. The system of claim 18, wherein the annotated question and the question that needs to be answered are an exact match.

20. The system of claim 18, wherein the parameterized question has an answer associated therewith, the question answering interface using the answer associated with the parameterized question to answer the question that needs to be answered.

21. The system of claim 18, wherein the question answering algorithm automatically extracts features from the question that needs to be answered, extracts a table name from the features, and extracts a parameterized question from the features.

22. The system of claim 18, wherein the question answering algorithm automatically extracts features from the search query, maps at least one of the features to a concept corresponding to the database;

and maps the question that needs to be answered to a matching tagged question.

23. A machine-readable medium that provides executable instructions which, when executed by a processor, cause the processor to perform a method for natural language searching comprising:

receiving a search query in the form of a natural language question;

normalizing the search query;

searching a database having a plurality of database items and questions tagged to the database associated with the database items for a question that matches the normalized search query; and

providing an answer corresponding to the database item and question that matches the normalized search query.

24. The machine-readable medium of claim 23, wherein the question is a parameterized question.

25. The machine-readable medium of claim 23, wherein the database items are arranged into a plurality of columns and each of the columns has a question tagged to the column.

26. The machine-readable medium of claim 25, wherein the question tagged to the column is a question answerable by the database items in the column.

27. The machine-readable medium of claim 23, wherein the search query and the tagged question are exact matches.

28. The machine-readable medium of claim 23, wherein the tagged question has the answer associated therewith.

29. The machine-readable medium of claim 28, wherein the answer is a parameterized answer, and at least one database item is substituted into the parameterized answer.

30. The machine-readable medium of claim 23, wherein normalizing the search query comprises one or more selected from the group consisting of removing unnecessary words from the search query, rephrasing the question into an answerable question, identifying and replacing words with synonyms, and modifying pluralities into singularities.

31. The machine-readable medium of claim 23, wherein the search query is matched to the tagged question by:

automatically extracting features from the search query;

identifying a database to be searched from the features; and

identifying a column with a tagged question from the features.

32. The machine-readable medium of claim 23, wherein the search query is matched to the tagged question by:

automatically extracting features from the search query;

mapping at least one of the features to a concept corresponding to the database; and

mapping the question that needs to be answered to a matching tagged question.

33. The machine-readable medium of claim 23, wherein the method further comprises identifying a database to search.

34. The machine-readable medium of claim 33, wherein identifying a database to search comprises comparing concepts extracted from the search query with concepts related to the database items in the database.

35. A database for answering questions comprising:

a plurality of database items organized into columns;

a plurality of questions, each of the plurality of questions annotated to a column, each of the questions being a parameterized question that is answerable by the database items in the column; and

a plurality of answers, each of the answers annotated to one of the plurality of questions, each of the answers being a parameterized answer that corresponds to the parameterized question, wherein at least one parameter in the parameterized answer are replaceable by at least one of the database items.

36. The database of claim 35, wherein the at least one parameter is replaceable by at least one of the database items in the column to which the parameterized answer is annotated.

37. The database of claim 35, wherein the parameterized answer includes a plurality of parameters, each of the parameters replaceable by a database item.

38. A method for natural language searching comprising:

receiving a search query in the form of a natural language question;

identifying a database that can provide an answer to the search query, the database having a plurality of database items organized according to database concepts;

searching the database for a database item that can answer the search query; and

providing an answer corresponding to the database item that answers search query.

39. The method of claim 38, further comprising identifying search query concepts of the search query and matching the search query concepts with the database concepts.

40. The method of claim 39, wherein matching the search query concepts with the database concepts comprises partially matching the search query concepts with the database concepts.

41. The method of claim 38, wherein statistical methods are used to provide the answer.

42. The method of claim 38, wherein string distance methods are used to provide the answer.

43. The method of claim 38, wherein the database does not include question tags or answer tags.