| 摘要: |
| 首先利用bidirectional encoder representations from transformers（BERT）模型的强大的语境理解能力来提取数据法律文本的深层语义特征,然后引入细粒度特征提取层,依照注意力机制,重点关注文本中与数据法律问答相关的关键部分,最后对所采集的法律问答数据集进行训练和评估. 结果显示：与传统的多个单一模型相比,所提出的模型在准确度、精确度、召回率、F1分数等关键性能指标上均有提升,表明该系统能够更有效地理解和回应复杂的数据法学问题,为研究数据法学的专业人士和公众用户提供更高质量的问答服务. |
| 关键词: bidirectional encoder representations from transformers(BERT)模型 细粒度特征提取 注意力机制 自然语言处理(NLP) |
| DOI：10.3969/J.ISSN.1000-5137.2024.02.010 |
| 分类号:TP3911 |
| 基金项目:上海市科学仪器领域项目（22142201900）；教育部重大项目（20JZD020）；国家自然科学基金（62301320） |
|
| Data law Q&A system based on BERT and fine-grained feature extraction |
|
SONG Wenhao1, WANG Yang1, ZHU Sulei1, ZHANG Qian1, WU Xiaoyan2
|
|
1.College of Information, Mechanical and Electrical Engineering, Shanghai Normal University, Shanghai 201418, China;2.School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
|
| Abstract: |
| A data legal question and answer system was proposed based on bidirectional encoder representations from transformers（BERT） model and fine-grained feature extraction to provide accurate professional legal consulting services. Firstly, the powerful contextual understanding ability of the BERT model was leveraged to extract deep semantic features from data legal texts. Subsequently, a fine-grained feature extraction layer was introduced which mainly focused on key components related to data legal Q&A within the text using an attention mechanism. Finally, the collected legal Q&A dataset was trained and evaluated. The results indicated that compared to traditional multiple single models, the proposed model showed improvements in key performance indicators such as accuracy, precision, recall, and F1 score, which suggested that the system could more effectively comprehend and address complex issues in data law, providing higher quality Q&A services for both research data law professionals and the general public. |
| Key words: bidirectional encoder representations from transformers(BERT) model fine-grained feature extraction attention mechanism natural language processing (NLP) |
