Semantic similarity model training method and device, semantic matching method and device
By utilizing iterative training of predicted semantic similarity and reference semantic similarity in the Sentence-BERT model, the problem of inconsistency between the model training process and the prediction process is solved, achieving a highly accurate semantic similarity matching model and improving the model's optimization efficiency and convergence speed.
CN115982597BActive Publication Date: 2026-05-26AVATR CO LTD
View PDF 2 Cites 0 Cited by
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- AVATR CO LTD
- Filing Date
- 2023-02-15
- Publication Date
- 2026-05-26
Smart Images
Figure CN115982597B_ABST
Abstract
This invention relates to the field of semantic recognition technology, and discloses a semantic similarity model training method and apparatus, and a semantic matching method and apparatus. The method includes: acquiring a training sample set, wherein each group of training samples in the training sample set includes a first sample text, a second sample text, and a reference semantic similarity between the first sample text and the second sample text; obtaining a predicted semantic similarity between the first sample text and the second sample text based on each group of training samples and an initial semantic similarity model; using the predicted semantic similarity as the initial training output of the initial semantic similarity model, and the reference semantic similarity as supervision information, iteratively training the initial semantic similarity model to obtain a target semantic similarity model. Applying the technical solution of this invention can solve the problem of inconsistency between the training and prediction stages in existing models, making the training results consistent with the prediction results.
Need to check novelty before this filing date? Find Prior Art