Abstract
知识库问答依靠知识库推断答案需大量带标注信息的问答对,但构建大规模且精准的数据集不仅代价昂贵,还受领域等因素限制。为缓解数据标注问题,面向知识库的问题生成任务引起了研究者关注,该任务是利用知识库三元组自动生成问题。现有方法仅由一个三元组生成的问题简短且缺乏多样性。为生成信息量丰富且多样化的问题,本文采用Graph Transformer和BERT两个编码层来加强三元组多粒度语义表征以获取背景信息。在SimpleQuestions上的实验结果证明了该方法有效性。- Anthology ID:
- 2020.ccl-1.31
- Volume:
- Proceedings of the 19th Chinese National Conference on Computational Linguistics
- Month:
- October
- Year:
- 2020
- Address:
- Haikou, China
- Venue:
- CCL
- SIG:
- Publisher:
- Chinese Information Processing Society of China
- Note:
- Pages:
- 324–335
- Language:
- Chinese
- URL:
- https://aclanthology.org/2020.ccl-1.31
- DOI:
- Cite (ACL):
- Yue Hu and Guangyou Zhou. 2020. 基于Graph Transformer的知识库问题生成(Question Generation from Knowledge Base with Graph Transformer). In Proceedings of the 19th Chinese National Conference on Computational Linguistics, pages 324–335, Haikou, China. Chinese Information Processing Society of China.
- Cite (Informal):
- 基于Graph Transformer的知识库问题生成(Question Generation from Knowledge Base with Graph Transformer) (Hu & Zhou, CCL 2020)
- PDF:
- https://preview.aclanthology.org/remove-xml-comments/2020.ccl-1.31.pdf