A framework for integrating nested queries in natural language interfaces to databases
Title
A framework for integrating nested queries in natural language interfaces to databases
Subject
Computer Science
Description
To translate Natural Language (NL) statements into Structured Query Language
(SQL) queries, different methods and systems were proposed in the past. This work presents a framework for automating the translation of Data Requirement
Specifications (DRS) given by enterprise Business Users in NL into SQL queries,
focusing on requirements that result in the generation of nested SQLs. The framework takes the business user’s DRS given in English as input and generates an initial query sketch by employing semantic parsing. This initial sketch is further refined and completed into a well-formed SQL by consulting the Database Schema. It performs the translation by combining NL processing techniques with Query Sketch generation methods and refines it by employing Repair techniques or extends it further. The framework suggests using Lambda expressions for intermediate representation and employs standard operations of Lambda Calculus for performing the required transformations needed for translation. Lambda Context Calculus (LCC) provides the operational semantics and the relevant methods needed in transforming NL statements into SQL, preserving the integrity and compositionality24 of expressions in every step of the translation. Though Lambda Calculus is found to be effective in representing the intermediate expressions and assists in performing the transformations that are needed for translating specific predicates into SQL, its inflexibility in combining parallel computations is a constraint. To represent clauses that are in parallel or are in
pipeline, and to perform the required transformations on the intermediate expressions involving these, more advanced programming constructs are needed. It also adopts functional programming techniques to deal with complex scenarios involving nested queries. This work recommends the use of some advanced language constructs, the Fixed-point Combinators11 and Monad Comprehensions20 for performing the required transformation at the intermediate language level and adopts functional programming techniques for the required syntactic sugar.
(SQL) queries, different methods and systems were proposed in the past. This work presents a framework for automating the translation of Data Requirement
Specifications (DRS) given by enterprise Business Users in NL into SQL queries,
focusing on requirements that result in the generation of nested SQLs. The framework takes the business user’s DRS given in English as input and generates an initial query sketch by employing semantic parsing. This initial sketch is further refined and completed into a well-formed SQL by consulting the Database Schema. It performs the translation by combining NL processing techniques with Query Sketch generation methods and refines it by employing Repair techniques or extends it further. The framework suggests using Lambda expressions for intermediate representation and employs standard operations of Lambda Calculus for performing the required transformations needed for translation. Lambda Context Calculus (LCC) provides the operational semantics and the relevant methods needed in transforming NL statements into SQL, preserving the integrity and compositionality24 of expressions in every step of the translation. Though Lambda Calculus is found to be effective in representing the intermediate expressions and assists in performing the transformations that are needed for translating specific predicates into SQL, its inflexibility in combining parallel computations is a constraint. To represent clauses that are in parallel or are in
pipeline, and to perform the required transformations on the intermediate expressions involving these, more advanced programming constructs are needed. It also adopts functional programming techniques to deal with complex scenarios involving nested queries. This work recommends the use of some advanced language constructs, the Fixed-point Combinators11 and Monad Comprehensions20 for performing the required transformation at the intermediate language level and adopts functional programming techniques for the required syntactic sugar.
Creator
George, Mathew - 1740071
Publisher
CHRIST (Deemed to be University)
Language
English
Type
PhD
Collection
Citation
George, Mathew - 1740071, “A framework for integrating nested queries in natural language interfaces to databases,” CHRIST (Deemed To Be University) Institutional Repository, accessed December 22, 2024, https://archives.christuniversity.in/items/show/1754.