Language Understanding Models Weighted by Human Perception
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional language understanding systems treat all tagged items equally during training, failing to account for human perception of their relative importance, which can lead to errors in intent and slot prediction that differ significantly in impact.
Innovation Solution
Incorporating human perception ratings to adjust the training of language understanding models by assigning weights to tagged items based on their perceived importance, using methods such as simple weight vectors or models trained with features, to prioritize items that are more critical to user satisfaction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional language understanding systems treat all tagged items equally during training, then the model training process is simple and uniform, but the model fails to account for human perception of relative importance leading to errors in intent and slot prediction
Solution Approach 1:
The patent applies local quality by assigning different weights to different tagged items (slots, intents, domains) based on their perceived importance. Instead of uniform treatment, each item receives a specific weight reflecting its relative importance to user satisfaction, allowing the model to focus more on critical items while maintaining manageable complexity through structured weighting mechanisms
2Manufacturing precision
If human perception ratings are incorporated to adjust training by assigning weights to tagged items, then the model better discriminates between errors and improves user experience, but the training process becomes more complex
Solution Approach 1:
The patent changes the parameter of item importance from uniform to variable by incorporating human perception ratings. These ratings transform the training process by adjusting weight parameters for different tagged items, enabling the model to discriminate between errors of varying importance. The complexity is managed through systematic parameter adjustment rather than fundamental process redesign
Data Source
AI summary
An understanding model is trained to account for human perception of the perceived relative importance of different tagged items (e.g. slot/intent/domain). Instead of treating each tagged item as equally important, human perception is used to adjust the training of the understanding model by associating a perceived weight with each of the different predicted items. The relative perceptual importance of the different items may be modeled using different methods (e.g. as a simple weight vector, a model trained using features (lexical, knowledge, slot type, . . . ), and the like). The perceptual weight vector and/or or model are incorporated into the understanding model training process where items that are perceptually more important are weighted more heavily as compared to the items that are determined by human perception as less important.