Database query language transformation method, transformation apparatus and database query system

• US 9,223,827 B2
• Filed: 04/03/2008
• Issued: 12/29/2015
• Est. Priority Date: 06/23/2006
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Hou, Xue Qiao, Hu, Gang, Liu, Tao, Qian, Qian
• Primary Examiner(s): Mamillapalli, Pavan
• Application Number:
  US12/061,742
• Publication Number:
  US 20090094216A1
• Time in Patent Office: 2,826 Days
• Field of Search: 707/713, 707/718, 707/719, 707/720
• US Class Current: 1/1
• CPC Class Codes: G06F 16/2452 Query translation
• CPC Class Codes: Y10S 707/99934 Query formulation, input preparation, or translation

Abstract

A method, device and system for transforming a concept-based query into SQL query statements, that includes transforming inputted concept-based queries into logic rules, checking validity of the logic rules, optimizing the valid logic rules, and translating the logic rules into SQL query statements in accordance with the optimization result.

Claims

1. A method, comprising:

transforming a concept-based query into logic rules, said transforming comprising concept to predicate transformation and property to variable transformation based on a pre-determined mapping table, said logic rules comprising an expression of a plurality of conditions and a conclusion, said conclusion being output when all of said plurality of conditions are established;

checking validity of said logic rules by determining whether said logic rules conform to pre-determined requirements and belong to a range of objects that a computing device can analyze and process, said checking validity of said logic rules comprising;

syntax checking whether said logic rules conform to defined syntax of a logic rule descriptive language; and

semantic checking whether said logic rules are logically valid;

for logic rules determined to be logically invalid, modifying said concept-based query to obtain valid logic rules;

optimizing logic rules that are valid by establishing a dependency model for relations between predicates in said logic rules, so that said logic rules can be translated into SQL query statements, said dependency model indicating an order for generating said SQL query statements; and

translating logic rules that have been optimized into said SQL query statements using a pre-determined mapping table, said pre-determined mapping table mapping predicates to relations in a relational database and mapping variables to entities in said relational database.

2. The method according to claim 1, said transforming a concept-based query into logic rules further comprising:

performing constraint normalization on said concept-based query, said constraint normalization comprising using all class object constraints as separate predicates and transforming all data type constraints into computing predicates, producing a constraint-normalized query statement;

performing conceptual model relational inference on said constraint-normalized query statement in accordance with a conceptual model;

performing transformation from concepts in said conceptual model to predicates in said logic rules and from properties in said conceptual model to variables in said logic rules based on a predetermined mapping table to obtain a transformed query statement; and

performing normal-form normalization on said transformed query statement to obtain said logic rules.

3. The method according to claim 2, said performing conceptual model relational inference on said constraint-normalized query statement further comprising performing subsumption inference on said constraint-normalized query statement;

and

said performing normal-form normalization on said transformed query statement to obtain said logic rules comprising performing disjunctive normal-form normalization on said transformed query statement.

4. The method according to claim 3, said checking validity of said logic rules further comprising:

syntax checking whether said logic rules that have undergone normal-form normalization conform to defined syntax of a logic rule descriptive language.

5. The method according to claim 4, said syntax checking further comprising safe rule checking of said logic rules, including checking for undefined predicates and undefined variables and consistency checking of said predicates and variables;

and

said semantic checking further comprising checking whether there are non-linear recursive predicates, checking whether there are indirect recursive predicates, and checking whether there is un-stratified negation.

6. The method according to claim 1, said optimizing further comprising:

building a dependency graph in accordance with dependency relations amongst predicates in said logic rules, every node corresponding to one predicate in said logic rules and dependency relations amongst respective predicates being denoted by directed edges amongst nodes; and

optimizing said logic rules in accordance with said dependency graph.

7. The method according to claim 6, said optimizing said logic rules in accordance with said dependency graph comprising:

generating an optimal processing order for predicates, in accordance with said dependency graph; and

pushing constraints in higher level predicates in said dependency graph downwards to lower level predicates in said dependency graph.

8. The method according to claim 3, said translating said logic rules into SQL query statements comprising:

translating every disjunction in said logic rules that have undergone disjunctive normal-form normalization into an SQL SELECT query statement; and

joining respective separate SQL SELECT query statements by using a keyword UNION ALL,

predicates in said logic rules being transformed into table names of a relational database, and variables in said logic rules being transformed into column names of said relational database, based on a predetermined mapping table.

9. The method according to claim 8, said translating every disjunction in said logic rules that have undergone disjunctive normal-form normalization into an SQL SELECT query statement comprising:

constructing a SELECT block of an SQL query statement by using column names corresponding to arguments in a head portion of said logic rules as attributes that follow a keyword SELECT;

constructing a FROM block of an SQL query statement by using table names corresponding to predicates involved in a body portion of said logic rules as table names that follow a keyword FROM; and

constructing a WHERE block of an SQL query statement by analyzing relations amongst variables of respective predicates in said logic rules and predicates expressing relation operations.

10. The method according to claim 8, a recursive logic rule being translated into an SQL query statement using a common table expression, a table name corresponding to a recursive predicate in said recursive logic rule being used as said table name in said common table expression.

11. A transformation apparatus, comprising:

a computing device, comprising;

a transformer transforming a concept-based query input to said computing device into logic rules, said transformer comprising a pre-determined mapping table, said transforming comprising concept to predicate transformation and property to variable transformation based on said pre-determined mapping table, said logic rules comprising an expression of a plurality of conditions and a conclusion, said conclusion being output when all said plurality of conditions are established;

a checker checking validity of said logic rules by determining whether said logic rules conform to pre-determined requirements and belong to a range of objects that said computing device can analyze and process, said checker comprising;

a syntax checker checking whether said logic rules conform to defined syntax of a defined logic rule descriptive language, and

a semantic checker checking whether said logic rule are logically valid,

said checker returning logic rules determined to be invalid in order to modify said concept-based query;

an optimizer optimizing said logic rules that have been checked to be valid by said checker, said optimizer establishing a dependency model for relations between predicates in said logic rules, so that said logic rules can be translated into SQL query statements, said dependency model indicating an order for generating said SQL query statements; and

a translator translating logic rules that have been optimized into said SQL query statements, said translator comprising a pre-determined mapping table, said pre-determined mapping table mapping predicates to relations in a relational database and mapping variables to entities in said relational database.

12. The transformation apparatus according to claim 11, further comprising a mapping table memory providing said transformer with concept-to-predicate and property-to-variable mapping information, and

said transformer:

performing constraint normalization on said concept-based query, said constraint normalization comprising using all class object constraints as separate predicates and transforming all data type constraints into computing predicates, producing a constraint-normalized query statement;

performing conceptual model relational inference on said constraint-normalized query statement in accordance with a conceptual model;

performing transformation from concepts in said conceptual model to predicates in said logic rules and from properties in said conceptual model to variables in said logic rules based on a predetermined mapping table to obtain a transformed query statement; and

performing normal-form normalization on said transformed query statement to obtain logic rules.

13. The transformation apparatus according to claim 12,

said computing device performing conceptual model relational inference on said constraint-normalized query statement comprising:

said computing device performing subsumption inference on said constraint-normalized query statement; and

said computing device performing normal-form normalization on said transformed query statement to obtain logic rules comprising;

said computing device performing disjunctive normal-form normalization on said transformed query statement.

14. The transformation apparatus according to claim 13, said syntax checker checking whether said logic rules that have undergone normal-form normalization conform to defined syntax of a defined logic rule descriptive language.

15. The transformation apparatus according to claim 14,

said syntax checker further performing safe rule checking on said logic rules, including checking for undefined predicates and undefined variables and consistency checking of said predicates and variables;

and

said semantic checker further checking whether there are non-linear recursive predicates, whether there are indirect recursive predicates, and whether there is unstratified negation.

16. The transformation apparatus according to claim 11, said optimizer:

building a dependency graph in accordance with dependency relations amongst predicates in said logic rules, wherein every node corresponds to one predicate in said logic rules and dependency relations amongst respective predicates are denoted by directed edges among nodes; and

optimizing said logic rules in accordance with said dependency graph.

17. The transformation apparatus according to claim 16, said optimizing said logic rules in accordance with said dependency graph comprising:

said computing device generating an optimal processing order for predicates, in accordance with said dependency graph; and

said computing device pushing constraints in higher level predicates of said dependency graph downwards to lower level predicates of said dependency graph.

18. The transformation apparatus according to claim 13, said mapping table memory further providing said translator with predicate-to-relation and variable-to-entity mapping information;

and

said translator;

translating every disjunction in said logic rules that have undergone disjunctive normal-form normalization into an SQL SELECT query statement; and

joining respective separate SQL SELECT query statements by a keyword UNION ALL,

predicates in said logic rules being transformed into table names of a relational database, and variables in said logic rules being transformed into column names of said relational database, based on a predetermined mapping table.

19. The transformation apparatus according to claim 18, said computing device translating every disjunction in said logic rules that have undergone disjunctive normal-form normalization into a SQL SELECT query statement comprising:

constructing a SELECT block of an SQL query statement by using column names corresponding to arguments in a head portion of said logic rules as attributes that follow a keyword SELECT;

constructing a FROM block of an SQL query statement by using table names corresponding to predicates involved in a body portion of said logic rules as table names that follow a keyword FROM; and

constructing a WHERE block of an SQL query statement by analyzing relations amongst variables of respective predicates in said logic rules and predicates expressing relation operations.

20. The transformation apparatus according to claim 18, said translator translating a recursive logic rule into an SQL query statement using a common table expression, a table name corresponding to a recursive predicate in said recursive logic rule being used as said table name in said common table expression.

21. A non-transitory computer program product containing computer program codes for implementing a method comprising:

transforming a concept-based query into logic rules, said transforming comprising concept to predicate transformation and property to variable transformation based on a pre-determined mapping table, said logic rules comprising an expression of a plurality of conditions and a conclusion, said conclusion being output when all said plurality of conditions are established;

checking validity of said logic rules by determining whether said logic rules conform to pre-determined requirements and belong to a range of objects that a computing device can analyze and process, said checking validity of said logic rules comprising;

syntax checking whether said logic rules conform to defined syntax of a logic rule descriptive language; and

semantic checking whether said logic rules are logically valid;

for logic rules determined to be logically invalid, modifying said concept-based query to obtain valid logic rules;

optimizing logic rules that are valid by establishing a dependency model for relations between predicates in said logic rules, so that said logic rules can be translated into SQL query statements, said dependency model indicating an order for generating said SQL statements; and

translating logic rules that have been optimized into said SQL query statements using a pre-determined mapping table, said pre-determined mapping table mapping predicates to relations in a relational database and mapping variables to entities in said relational database.