Mapping and boosting of terms in a format independent data retrieval query

• US 9,069,882 B2
• Filed: 01/22/2013
• Issued: 06/30/2015
• Est. Priority Date: 01/22/2013
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Diament, Judah M., Heching, Aliza R.
• Primary Examiner(s): Stevens, Robert
• Application Number:
  US13/746,440
• Publication Number:
  US 20140207790A1
• Time in Patent Office: 889 Days
• Field of Search: 707/748
• US Class Current: 1/1
• CPC Class Codes: G06F 16/90335 Query processing

Abstract

Aspects of the present invention provide a tool for retrieving data in response to a query in any format. In an embodiment, each query term of the query is analyzed to determine whether it corresponds to either a domain keyword or a formula designator. If a formula designator is retrieved, each formula term associated with the formula designator is mapped to a domain keyword. Each of the domain keywords, the formula designators, and the mapped formula terms are boosted. A dataset is searched using a structured search query that includes all boosted terms, as well as any remaining query terms that do not correspond to one of the boosted terms.

Claims

1. A method for retrieving data, comprising:

analyzing each query term of a format-independent query for data retrieval to determine whether the query term corresponds to at least one of;

a domain keyword or a formula designator;

wherein in the case that at least one domain keyword is retrieved as corresponding to the query term, mapping the retrieved domain keyword to the query term such that the at least one retrieved domain keyword is substituted for the query term in the format-independent query,

wherein in the case that at least one formula designator is retrieved as corresponding to the query term, mapping a set of formula terms associated with the retrieved formula designator to the query term such that the at least one retrieved formula designator is substituted for the query term in the format-independent query,

wherein in the case that both at least one domain keyword and at least one formula designator are retrieved as corresponding to the query term, mapping the set of formula terms associated with the at least one of the retrieved formula designators to the at least one of the retrieved domain keywords of the query term;

creating a unique weighted boosting for each of the retrieved domain keywords, the retrieved formula designators, and the mapped formula terms; and

searching a dataset using a structured search query that includes all weighted boosted terms and any remaining query terms that do not correspond to one of the weighted boosted terms.

2. The method of claim 1, in the case that both the at least one domain keyword and the at least one formula designator are retrieved as corresponding to the query term, the mapping further comprising:

evaluating a formula definition associated with the retrieved formula designator to generate a set of formula terms;

evaluating domain keywords corresponding to the formula designator to determine whether a corresponding domain keyword corresponds to one of the set of formula terms;

mapping the query terms to each of the corresponding formula terms;

determining whether all of the set of formula terms have been mapped; and

based on a determination that all of the set of formula terms have not been mapped, inserting an unmapped formula term from the set of formula terms into the query.

3. The method of claim 1, wherein the query is in a natural language format.

4. The method of claim 1, the analyzing further comprising:

searching a business domain vocabulary to determine whether the query term corresponds to a domain keyword; and

searching at least one of the business domain vocabulary or a statistical vocabulary to determine whether the query term corresponds to a formula designator and returning a structural mapping that includes one or more constituent formula terms if a match is found.

5. The method of claim 1, the searching further comprising:

accessing a dataset description of a plurality of datasets, the dataset description including data descriptors that describe a value set in the plurality of datasets; and

comparing the boosted terms and the remaining query items with the data descriptors;

evaluating value sets associated with domain keywords associated with a particular formula designator;

boosting a returned value set based on a designation type of the data descriptor corresponding to the returned value set, the boosting assigning a first boost value to any data descriptors having a preferred designation type that is greater than a second boost value assigned to data descriptors having an inferior designation type;

ranking a solution set based on the boosted terms and the boosted return value set; and

returning the ranked solution set.

6. The method of claim 5, wherein the boosting determines which designation type of a set of designation types has a preferred designation type based on at least one of:

a user identity, a user position, a user group, a software application used, or a point in time, wherein the set of designation types include;

key name, column name, column description, value description, table name, and table description.

7. The method of claim 5, further comprising:

linking a first returned value set having a particular data descriptor with a second returned value set having a corresponding data descriptor; and

returning the linking and the data descriptors with the ranked solution set.

8. The method of claim 5, further comprising providing a graphical user interface that enables a user to add, delete, and modify data descriptors in the dataset description.

9. A system for retrieving data, comprising at least one computer device that performs a method, comprising:

analyzing each query term of a format-independent query for data retrieval to determine whether the query term corresponds to at least one of;

a domain keyword or a formula designator;

wherein in the case that at least one domain keyword is retrieved as corresponding to the query term, mapping the retrieved domain keyword to the query term such that the at least one retrieved domain keyword is substituted for the query term in the format-independent query,

wherein in the case that at least one formula designator is retrieved as corresponding to the query term, mapping a set of formula terms associated with the retrieved formula designator to the query term such that the at least one retrieved formula designator is substituted for the query term in the format-independent query,

wherein in the case that both at least one domain keyword and at least one formula designator are retrieved as corresponding to the query term, mapping the set of formula terms associated with the at least one of the retrieved formula designators to the at least one of the retrieved domain keywords of the query term;

creating a unique weighted boosting for each of the retrieved domain keywords, the retrieved formula designators, and the mapped formula terms; and

searching a dataset using a structured search query that includes all weighted boosted terms and any remaining query terms that do not correspond to one of the weighted boosted terms.

10. The system of claim 9, in the case that both the at least one domain keyword and the at least one formula designator are retrieved as corresponding to the query term, the mapping further comprising:

evaluating a formula definition associated with the retrieved formula designator to generate a set of formula terms;

evaluating domain keywords corresponding to the formula designator to determine whether a corresponding domain keyword corresponds to one of the set of formula terms;

mapping the query terms to each of the corresponding formula terms;

determining whether all of the set of formula terms have been mapped; and

based on a determination that all of the set of formula terms have not been mapped, inserting an unmapped formula term from the set of formula terms into the query.

11. The system of claim 9, wherein the query is in a natural language format.

12. The system of claim 9, the analyzing further comprising:

searching a business domain vocabulary to determine whether the query term corresponds to a domain keyword; and

searching at least one of the business domain vocabulary or a statistical vocabulary to determine whether the query term corresponds to a formula designator and returning a structural mapping that includes one or more constituent formula terms if a match is found.

13. The system of claim 9, the searching further comprising:

accessing a dataset description of a plurality of datasets, the dataset description including data descriptors that describe a value set in the plurality of datasets; and

comparing the boosted terms and the remaining query items with the data descriptors;

evaluating value sets associated with domain keywords associated with a particular formula designator;

boosting a returned value set based on a designation type of the data descriptor corresponding to the returned value set, the boosting assigning a first boost value to any data descriptors having a preferred designation type that is greater than a second boost value assigned to data descriptors having an inferior designation type;

ranking a solution set based on the boosted terms and the boosted return value set; and

returning the ranked solution set.

14. The system of claim 13, wherein the boosting determines which designation type of a set of designation types has a preferred designation type based on at least one of:

a user identity, a user position, a user group, a software application used, or a point in time, wherein the set of designation types include;

key name, column name, column description, value description, table name, and table description.

15. The system of claim 13, the method further comprising:

linking a first returned value set having a particular data descriptor with a second returned value set having a corresponding data descriptor; and

returning the linking and the data descriptors with the ranked solution set.

16. The system of claim 13, the method further comprising providing a graphical user interface that enables a user to add, delete, and modify data descriptors in the dataset description.

17. A computer program product stored on a non-transitory computer readable storage medium, which, when executed performs a method for retrieving data, comprising:

analyzing each query term of a format-independent query for data retrieval to determine whether the query term corresponds to at least one of;

a domain keyword or a formula designator;

wherein in the case that at least one domain keyword is retrieved as corresponding to the query term, mapping the retrieved domain keyword to the query term such that the at least one retrieved domain keyword is substituted for the query term in the format-independent query,

wherein in the case that at least one formula designator is retrieved as corresponding to the query term, mapping a set of formula terms associated with the retrieved formula designator to the query term such that the at least one retrieved formula designator is substituted for the query term in the format-independent query,

wherein in the case that both at least one domain keyword and at least one formula designator are retrieved as corresponding to the query term, mapping the set of formula terms associated with the at least one of the retrieved formula designators to the at least one of the retrieved domain keywords of the query term;

creating a unique weighted boosting for each of the retrieved domain keywords, the retrieved formula designators, and the mapped formula terms; and

searching a dataset using a structured search query that includes all weighted boosted terms and any remaining query terms that do not correspond to one of the weighted boosted terms.

18. The program product of claim 17, in the case that both the at least one domain keyword and the at least one formula designator are retrieved as corresponding to the query term, the mapping further comprising:

evaluating a formula definition associated with the retrieved formula designator to generate a set of formula terms;

evaluating domain keywords corresponding to the formula designator to determine whether a corresponding domain keyword corresponds to one of the set of formula terms;

mapping the query terms to each of the corresponding formula terms;

determining whether all of the set of formula terms have been mapped; and

based on a determination that all of the set of formula terms have not been mapped, inserting an unmapped formula term from the set of formula terms into the query.

19. The program product of claim 17, wherein the query is in a natural language format.

20. The program product of claim 17, the analyzing further comprising:

searching a business domain vocabulary to determine whether the query term corresponds to a domain keyword; and

searching at least one of the business domain vocabulary or a statistical vocabulary to determine whether the query term corresponds to a formula designator and returning a structural mapping that includes one or more constituent formula terms if a match is found.

21. The program product of claim 17, the searching further comprising:

accessing a dataset description of a plurality of datasets, the dataset description including data descriptors that describe a value set in the plurality of datasets; and

comparing the boosted terms and the remaining query items with the data descriptors;

evaluating value sets associated with domain keywords associated with a particular formula designator;

boosting a returned value set based on a designation type of the data descriptor corresponding to the returned value set, the boosting assigning a first boost value to any data descriptors having a preferred designation type that is greater than a second boost value assigned to data descriptors having an inferior designation type;

ranking a solution set based on the boosted terms and the boosted return value set; and

returning the ranked solution set.

22. The program product of claim 21, wherein the boosting determines which designation type of a set of designation types has a preferred designation type based on at least one of:

a user identity, a user position, a user group, a software application used, or a point in time, wherein the set of designation types include;

key name, column name, column description, value description, table name, and table description.

23. The program product of claim 21, the method further comprising:

linking a first returned value set having a particular data descriptor with a second returned value set having a corresponding data descriptor; and

returning the linking and the data descriptors with the ranked solution set.

24. The program product of claim 21, the method further comprising providing a graphical user interface that enables a user to add, delete, and modify data descriptors in the dataset description.

25. A method for deploying an application for retrieving data, comprising:

providing a computer infrastructure being configured to;

analyze each query term of a format-independent query for data retrieval to determine whether the query term corresponds to at least one of;

a domain keyword or a formula designator;

wherein in the case that at least one domain keyword is retrieved as corresponding to the query term, map the retrieved domain keyword to the query term such that the at least one retrieved domain keyword is substituted for the query term in the format-independent query,

wherein in the case that at least one formula designator is retrieved as corresponding to the query term, map a set of formula terms associated with the retrieved formula designator to the query term such that the at least one retrieved formula designator is substituted for the query term in the format-independent query,

wherein in the case that both at least one domain keyword and at least one formula designator are retrieved as corresponding to the query term, map the set of formula terms associated with the at least one of the retrieved formula designators to the at least one of the retrieved domain keywords of the query term;

create a unique weighted boosting for each of the retrieved domain keywords, the retrieved formula designator, and the mapped formula terms; and

search a dataset using a structured search query that includes all weighted boosted terms and any remaining query terms that do not correspond to one of the weighted boosted terms.