SQL Syntax Extension for Machine Learning Service Invocation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users need to be proficient in both SQL and non-SQL languages like Python to perform machine learning model training using SQL statements, imposing higher demands on users and complicating the process.
Innovation Solution
Extending existing SQL syntax to include information about computing elements needed for a computing service, allowing for the generation of a computing service description statement that can be recognized by a target computing framework, thereby enabling computation without requiring users to write programs specific to that framework.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If users write machine learning scripts using R or Python and embed them in SQL programs, then computing services can be implemented, but the user needs to be familiar with multiple programming languages and write complex programs
Solution Approach 1:
The patent introduces an intermediary translation layer that converts SQL statements with computing service requirements into executable programs for the target computing framework. This intermediary automatically handles the complexity of program writing, allowing users to simply write SQL statements while the system translates them into the appropriate computing framework code, thus resolving the contradiction between implementing computing services and reducing programming requirements.
Solution Approach 2:
The patent creates a universal SQL-based interface that can invoke computing services across different computing frameworks (TensorFlow, PyTorch, etc.) without requiring users to learn framework-specific programming. The extended SQL syntax serves as a universal language that adapts to multiple computing backends, eliminating the need for users to write separate programs for each framework while maintaining broad computing service compatibility.
2Ease of operation
If existing SQL syntax is used without extension, then SQL statements remain simple, but they cannot include information about computing elements needed by computing services
Solution Approach 1:
The patent segments the SQL statement into distinct functional components: data query parts and computing service specification parts. By introducing specific keywords (TRAIN, PREDICT, CLUSTER) and computing element clauses, the SQL statement is divided into manageable segments that can be independently parsed and processed. This segmentation allows the statement to remain syntactically simple while incorporating rich computing service information through structured extensions.
Solution Approach 2:
The patent adds a new dimension to SQL statements by incorporating computing service type and computing element information as additional layers of specification. Rather than extending the core SQL syntax complexity, it adds orthogonal dimensions (computing service type, computing elements) that can be appended to standard SQL queries, thereby maintaining simplicity in the base language while enabling comprehensive computing service capabilities.
Data Source
AI summary
Syntax parsing on a SQL statement is performed to determine whether an extended syntax identifier exists in the SQL statement, where the extended syntax identifier indicates a target computing service for the SQL statement. It is determined that the extended syntax identifier exists in the SQL statement. A computing service description statement in a first statement format is generated based on the SQL statement, where the first statement format is a statement format that can be recognized by a target computing framework. The computing service description statement is submitted to the target computing framework. Data queried by the SQL statement is invoked, in the target computing framework based on the computing service description statement, to perform target computation, where the SQL statement includes a computing element needed by the target computing service.