Dialog Intent Classification via Variant Assessment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cognitive systems, such as AI platforms, are inherently non-deterministic, leading to inconsistent data extraction and incorrect outputs due to the susceptibility of machine learning models to input variations and errors, which necessitates the creation of deterministic behavior.
Innovation Solution
A system that utilizes a knowledge engine and classifier to assess dialog input data, identify tokens, create alternative dialog inputs, and statistically validate classifications to ensure accurate output, transforming the AI platform to a deterministic state.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If machine learning models are used to process natural language in cognitive systems, then the system can learn from data and make decisions, but the system becomes non-deterministic and produces inconsistent outputs
Solution Approach 1:
The system performs preliminary actions by generating multiple alternative interpretations of input data before making a final decision. The hypothesis generator creates several possible entity extractions and classifications, which are then evaluated through cross-validation to determine the most accurate output, thereby achieving deterministic behavior while maintaining learning capabilities
Solution Approach 2:
The system implements feedback mechanisms through cross-validation where multiple machine learning models evaluate each other's outputs. The validation process provides feedback on the reliability of entity extractions, allowing the system to identify and correct inconsistent results, thus improving output consistency while preserving adaptability
2Productivity
If machine learning models are deployed to extract entities from data, then the system can process and analyze information, but errors in input documents lead to incorrect data extraction and output
Solution Approach 1:
The system applies local quality by treating different portions of the input data with different levels of scrutiny. Critical entities and relationships receive more rigorous validation through multiple model assessments, while less critical information undergoes standard processing. This selective approach maintains high accuracy for important data while preserving overall processing efficiency
Solution Approach 2:
The system performs preliminary validation by generating multiple alternative entity extractions before finalizing the output. By creating several hypotheses about entity identities and relationships, then cross-validating them against the input document and each other, the system identifies the most accurate extraction while filtering out errors from incorrect interpretations
Data Source
AI summary
A system, computer program product, and method are provided for use with an intelligent computer platform to identify intent and convert the intent to one or more physical actions. The aspect of converting intent includes receiving content, identifying potential variants, and statistically analyzing the variants with a confidence assessment. The variants are sorted based on a protocol associated with the confidence assessment. A variant from the sort is selected and applied to a physical device, which performs a physical action and an associated hardware transformation based on the variant.