Intention Recognition Model Training via Satisfaction Feedback
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In natural language processing, deep learning methods for intention recognition require a large number of samples for accurate model training, making it challenging to construct effective training datasets.
Innovation Solution
A method that acquires and filters dialogue data based on user satisfaction degrees to select target data for training, forming a closed loop from online recognition to offline training, continuously improving the accuracy of the intention recognition model.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a deep learning method based on neural networks is used for intention recognition, then the fitting ability and generalization ability are improved, but a large number of samples are needed to train the model, making sample construction difficult
Solution Approach 1:
The patent implements a feedback mechanism where the intention recognition model processes online dialogue data, and the results are fed back to construct training samples. The system collects dialogue data, performs intention recognition, evaluates the results, and uses high-quality recognized samples to retrain the model, creating a closed-loop feedback system that continuously improves model performance without requiring external large-scale sample collections
Solution Approach 2:
The system enables self-service by allowing the intention recognition model to generate its own training samples from online dialogue data. The model processes real user inputs, identifies high-quality recognition cases, and automatically constructs training datasets for self-retraining, making the system self-sufficient in sample generation without external intervention or pre-collected large datasets
2Reliability
If manual construction of training samples is performed to ensure model accuracy, then the intention recognition accuracy is improved, but the work量和 time consumption increase significantly
Solution Approach 1:
The system automatically constructs training samples through self-service mechanisms. The intention recognition model processes online dialogue data, evaluates recognition quality, and automatically selects high-quality samples for training. This automated sample construction process eliminates manual intervention, significantly reducing the time and labor required compared to traditional manual sample construction methods
Solution Approach 2:
The feedback mechanism enables automatic sample construction by continuously evaluating recognition results and feeding back high-quality samples to the training process. The system collects dialogue data, performs recognition, evaluates results against satisfaction conditions, and automatically generates training samples based on recognized high-quality cases, creating an efficient automated sample construction pipeline
Data Source
AI summary
A method and apparatus for recognizing a user intention, a device, and a readable storage medium are disclosed. A specific implementation of the method includes: acquiring a plurality of rounds of dialogue data and a satisfaction degree of each round of dialogue data, selecting target dialogue data having a satisfaction degree meeting a set satisfaction condition; labeling input data in the target dialogue data with an intention label; and training the intention recognition model based on the input data in the target dialogue data and the intention label of the input data, so that the trained intention recognition model performs intention recognition on new input data.