Semantic Mining Framework for Intent Entity Model Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing natural language understanding (NLU) systems require manual creation and frequent updates of intent models, which are costly and time-consuming, and struggle with accurately classifying intents from natural language utterances due to the subjective nature of intent expression and hierarchical relationships within conversational data.
Innovation Solution
A semantic mining framework that generates intent/entity models unsupervisedly from a corpus of source data, such as chat logs or email strings, by producing intent vectors and calculating distances to identify meaning clusters, allowing for automatic generation and improvement of intent models with minimal human intervention, and enabling the identification of intents and responses in virtual agents.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual creation and updates of intent models are used, then the system can accurately classify intents, but the time and cost for model creation and maintenance increase significantly
Solution Approach 1:
The system automatically mines intents from conversational data and generates intent models without human intervention. The intent mining process autonomously extracts intents, generates descriptions, and creates classification rules, eliminating the need for manual model creation and updates while maintaining high classification accuracy
Solution Approach 2:
The system performs preliminary intent mining on conversational data to pre-generate intent models before they are needed for classification. By proactively extracting intents and creating model structures in advance, the system reduces the time required for model creation when classification tasks arise
2Measurement precision
If manual creation and updates of intent models are used, then the system can classify intents, but the cost of model creation and maintenance increases
Solution Approach 1:
The system autonomously performs intent extraction, description generation, and model updates without requiring human expertise or manual intervention. This self-service capability eliminates labor costs associated with manual intent model creation while maintaining high classification accuracy through automated semantic analysis
Solution Approach 2:
The system replaces manual mechanical processes of intent model creation with automated computational processes. Machine learning algorithms and natural language processing automatically extract intents and generate classifications, substituting human expert labor with scalable computational methods that reduce costs
3Productivity
If automated intent mining is implemented, then the time and cost for model creation is reduced, but the complexity of the system increases
Solution Approach 1:
The system introduces an intermediary intent mining layer between raw conversational data and the intent classification model. This intermediary automatically extracts intents and generates descriptions, serving as a bridge that simplifies the overall process while managing complexity through modular architecture
Solution Approach 2:
The intent mining system is divided into separate functional modules: intent extraction, description generation, and model updating. This segmentation allows each component to handle specific tasks independently, reducing overall system complexity while improving creation efficiency through specialized processing
4Ease of operation
If traditional NLU systems are used, then intent classification can be performed, but the system struggles with subjective intent expression and hierarchical relationships
Solution Approach 1:
The system transforms subjective intent expressions into structured parameters through automated mining. By extracting intents from diverse conversational data and representing them as standardized parameters with hierarchical relationships, the system maintains ease of operation while improving classification accuracy through consistent parameterization
Data Source
AI summary
An agent automation system includes a memory configured to store a corpus of utterances and a semantic mining framework and a processor configured to execute instructions of the semantic mining framework to cause the agent automation system to perform actions, wherein the actions include: detecting intents within the corpus of utterances; producing intent vectors for the intents within the corpus; calculating distances between the intent vectors; generating meaning clusters of intent vectors based on the distances; detecting stable ranges of cluster radius values for the meaning clusters; and generating an intent/entity model from the meaning clusters and the stable ranges of cluster radius values, wherein the agent automation system is configured to use the intent/entity model to classify intents in received natural language requests.