Natural Language to SQL Schema Linking for Accurate Query Translation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional NL2SQL translators struggle with accurately correlating entities in user natural language queries with database features due to diversity and ambiguity, leading to reduced performance and slow responses.
Innovation Solution
Preprocessing techniques using named entity recognition and scalable searches for content-based schema linking to match utterance words with database entities, generating a data structure for improved natural language to logical form model response time and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional NL2SQL translators are used to translate natural language queries to SQL, then the system can handle basic queries, but the accuracy of correlating entities with database features deteriorates due to diversity and ambiguity in natural language
Solution Approach 1:
The patent introduces an intermediary preprocessing system that includes named entity recognition and content-based schema linking. This intermediary layer processes natural language queries before they reach the NL2SQL translator, extracting and standardizing entities to improve correlation accuracy with database features while handling diversity and ambiguity in the input language.
Solution Approach 2:
The patent applies preliminary action by performing named entity recognition and schema linking operations before the main translation process. This preprocessing step prepares the natural language query by identifying and categorizing entities in advance, making the subsequent translation more accurate and efficient.
2Productivity
If conventional NL2SQL translation methods are used, then the system can process queries, but the response time deteriorates due to slow entity correlation and schema linking
Solution Approach 1:
The patent performs schema linking and entity extraction in advance as a preprocessing step. By preparing the mapping between natural language entities and database features before the actual translation occurs, the system reduces the time required during query processing, thereby improving productivity while minimizing time loss.
Solution Approach 2:
The patent segments the query processing into distinct phases: named entity recognition, content-based schema linking, and SQL translation. This segmentation allows each component to be optimized independently and processed in parallel where possible, improving overall response time and productivity.
3Ease of manufacture
If rule-based systems are used for natural language to SQL translation, then the system is easier to implement, but the ability to generalize to new use cases deteriorates
Solution Approach 1:
The patent replaces rigid rule-based systems with machine learning-based named entity recognition and content-based schema linking. This substitution maintains ease of implementation through automated training while dramatically improving adaptability to new domains and use cases by learning from data rather than relying on hand-crafted rules.
4Adaptability or versatility
If deep learning models are used for natural language to SQL translation, then the system generalizes well to new use cases, but the complexity of the system deteriorates
Solution Approach 1:
The patent segments the complex deep learning system into separate, specialized components: a named entity recognition model and a content-based schema linking model. This segmentation reduces overall system complexity by allowing each component to be trained and optimized independently while maintaining strong generalization capabilities through their collaborative architecture.
Data Source
AI summary
Techniques for preprocessing data assets to be used in a natural language to logical form model based on scalable search and content-based schema linking. In one particular aspect, a method includes accessing an utterance, classifying named entities within the utterance into predefined classes, searching value lists within the database schema using tokens from the utterance to identify and output value matches including: (i) any value within the value lists that matches a token from the utterance and (ii) any attribute associated with a matching value, generating a data structure by organizing and storing: (i) each of the named entities and an assigned class for each of the named entities, (ii) each of the value matches and the token matching each of the value matches, and (iii) the utterance, in a predefined format for the data structure, and outputting the data structure.