Voice Assistant Intent Classification Using Ambiguity Verification Modules
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current electronic devices with voice assistant functions face challenges in accurately interpreting user utterances due to the need for large volumes of training data for neural network-based models, leading to inefficiencies in intent classification and ambiguity handling.
Innovation Solution
The implementation of an electronic device with a processor configured to convert user utterances into text data and utilize multiple ambiguity verification modules, including language models, to obtain ambiguity index values and provide responses based on these values, enabling efficient ambiguity detection and intent classification.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a neural network-based model is used for intent classification, then the accuracy of user intent recognition is improved, but the volume of training data required increases significantly
Solution Approach 1:
The patent segments the intent classification task into multiple stages: first using a language model to generate candidate intents, then applying ambiguity verification modules to evaluate and rank these candidates. This segmentation allows the system to achieve high accuracy without requiring the entire classification process to be trained on massive datasets simultaneously, thereby reducing the training data volume requirement while maintaining recognition precision.
Solution Approach 2:
The patent introduces ambiguity verification modules as intermediary components between the language model and the final intent determination. These modules act as mediators that evaluate the confidence levels of candidate intents generated by the language model and apply additional verification rules, enabling the system to achieve accurate intent recognition with less training data by leveraging intermediate verification steps.
2Measurement precision
If multiple ambiguity verification modules are used, then the accuracy of ambiguous utterance handling is improved, but the device complexity increases
Solution Approach 1:
The patent implements a dynamic ambiguity verification mechanism where the system adaptively selects and applies verification modules based on the specific characteristics of each utterance. Rather than rigidly applying all verification modules to every input, the system dynamically determines which modules are necessary for each case, thereby improving ambiguous utterance handling accuracy while avoiding unnecessary complexity in the overall system structure.
Solution Approach 2:
The patent applies different verification strategies and modules to different parts of the intent classification process based on local requirements. Specific verification modules are tailored to handle particular types of ambiguity (e.g., semantic ambiguity, syntactic ambiguity), allowing the system to achieve high accuracy in handling ambiguous utterances by applying targeted verification where needed rather than uniformly across all cases, thus managing complexity effectively.
Data Source
AI summary
An electronic device includes a memory including instructions and a processor electrically connected to the memory and configured to execute the instructions. When the instructions are executed by the processor, the processor is configured to convert an utterance into text data. The processor is configured to obtain a plurality of ambiguity index values from the text data based on each of a plurality of ambiguity verification modules including a language model. The processor is configured to provide a response to a user based on the plurality of ambiguity index values.