文本隐式篇章关系识别方法、系统、设备及存储介质
By combining RoBERTa and K-BERT models with Bi-LSTM and convolutional neural networks, the complexity of implicit discourse relation recognition in educational texts was addressed, achieving accurate understanding of professional terminology and multidisciplinary knowledge, and improving the recognition effect of implicit discourse relations.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- XI AN JIAOTONG UNIV
- Filing Date
- 2023-10-11
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies are not very effective in identifying implicit textual relationships in educational texts, especially when dealing with complex technical terms and multidisciplinary knowledge, making it difficult to accurately identify and understand implicit relationships in the text.
The RoBERTa pre-trained model is used for encoding, and knowledge graph information is introduced by combining the K-BERT model. The semantic vectors of arguments are obtained through the Bi-LSTM model. Following the human understanding process, fine-grained cue scores are calculated, a multi-angle cue matrix is constructed, features are extracted using a convolutional neural network, and finally implicit text relationship categories are output through the softmax function.
It improves the recognition of implicit discourse relationships in educational texts, better understands professional terms and complex terminology, achieves deeper argument interaction representation, and enhances the model's performance in the education field.
Smart Images
Figure CN117390186B_ABST