Schema-Driven Conversational Training Data for Intent Classification

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing conversational AI systems struggle to accurately understand user intents due to manual creation of domain-specific intents and predefined responses, limiting their ability to provide relevant information.

Innovation Solution

A method and system that utilizes structured database schemas to generate training data for conversational systems, employing SQL and NoSQL queries, Deep Learning Neural Networks, and semantic analysis to create a knowledge corpus for intent classification, enabling dynamic understanding of user queries.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual creation of domain-specific intents is used for training conversational AI, then the system can be trained with structured intent definitions, but the system fails to accurately understand user intents and is limited to predefined domains

Engineering Contradiction:

Improveintent classification accuracyVSAvoiddomain coverage

Core Design Contradiction:

Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The system automatically generates training data by converting database schemas to natural language queries using DNN models, eliminating the need for manual intent creation. This self-service approach enables the conversational AI to understand user intents across multiple domains without human intervention, simultaneously improving accuracy and domain versatility

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system transforms the training data generation process by changing from manual parameter definition to automated DNN-based generation. By using deep learning models to convert database schemas into natural language queries, the system dynamically adapts to different domains and improves intent classification accuracy through learned patterns rather than fixed manual definitions

Inventive Principle:

Principle #35Parameter changes

2Stability of the object's composition

If predefined templates and responses are used in conversational AI, then the system can provide consistent responses, but the system fails to provide relevant information for user queries

Engineering Contradiction:

Improveresponse consistencyVSAvoidquery understanding accuracy

Core Design Contradiction:

Stability of the object's compositionVSLoss of information

Solution Approach 1:

The system performs preliminary action by pre-generating comprehensive training data from database schemas before the conversational AI needs to respond to user queries. This advance preparation enables the system to maintain response consistency while improving query understanding, as the DNN models have already learned the relationship between database structures and natural language queries

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system introduces an intermediary layer of automatically generated training data that bridges the gap between rigid predefined templates and flexible query understanding. This intermediary training corpus, created through DNN-based conversion of database schemas, enables the conversational AI to maintain consistency while accurately understanding diverse user queries

Inventive Principle:

Principle #24Intermediary (Mediator)

3Productivity

If manual intent creation is used for training, then the training process can be controlled, but the productivity of training data generation is low and limited to few domains

Engineering Contradiction:

Improvetraining data generation speedVSAvoidnumber of domains

Core Design Contradiction:

ProductivityVSAdaptability or versatility

Solution Approach 1:

The system enables self-service training data generation where database schemas are automatically converted to natural language queries using DNN models. This eliminates the need for manual intent creation across multiple domains, simultaneously boosting productivity and expanding domain versatility without human intervention

Inventive Principle:

Principle #25Self-service

Solution Approach 2:

The system achieves universality by creating a single automated training data generation process that works across multiple domains. The DNN-based conversion of database schemas to natural language queries is domain-agnostic, enabling high-productivity training data generation for diverse domains through one unified approach

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS12405976B2Method and system for generating training data for classifying intents in a conversational system

Publication Date: 2025.09.02 HITACHI LTD

AI summary

Embodiments of present disclosure relates to method training data generation system for generating training data for classifying intent in conversational system. The training data generation system receives database schema and creates SQL/NoSQL queries. The training data generation system generates natural language queries for the SQL/NoSQL queries. Further, the training data generation system generates training data for intents associated with the natural language queries and provides to classification models associated with conversational system for classification of intents. Embodiments of present disclosure relates to method and conversational system for providing natural language response for query. The conversational system receives query from user and classifies intent of the query and provides relevant response by mapping the query with the SQL/NoSQL queries generated by the training data generation system. Thus, the present disclosure generates conversational system without manually providing training data for classifying intents in real-time.