Sample processing method and device, and identification model training method and device
By constructing phrase feature labels and calculating sample similarity to filter negative samples, the problems of overfitting and inaccurate negative sample labeling in the training of named entity recognition models are solved, thereby improving the recognition accuracy and efficiency of the model.
CN115687917BActive Publication Date: 2026-06-26ALIBABA (CHINA) CO LTD +1
3 Cites 0 Cited by
Patent Information
- Application Number
- CN202211289712.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-10-20
- Publication Date
- 2026-06-26
- Estimated Expiration
- 2042-10-20
AI Technical Summary
Technical Problem
Existing named entity recognition models suffer from overfitting during training, resulting in poor performance on real data. Furthermore, negative sample labeling is costly and its accuracy is difficult to guarantee.
Method used
By constructing phrases of the sentences to be processed and determining the feature labels of the word units, the similarity between the target positive samples and negative samples is calculated. Accurately labeled negative samples are selected and combined with the target positive samples for model training to avoid model overfitting.
Benefits of technology
It improves the accuracy of negative sample labeling, reduces manpower and time costs, enhances the model's recognition accuracy, and avoids model overfitting.
✦ Generated by Eureka AI based on patent content.
Abstract
Embodiments of the present specification provide a sample processing method and device, wherein the sample processing method comprises: obtaining a to-be-processed sentence; constructing a word group corresponding to the to-be-processed sentence, and determining a feature label corresponding to each word unit in the word group; constructing at least one negative sample and a target positive sample according to the feature label and the word group; calculating a sample similarity between the target positive sample and each negative sample, and determining a target negative sample in the at least one negative sample according to the sample similarity.
Need to check novelty before this filing date? Find Prior Art