AntNLP at CoNLL 2018 Shared Task: A Graph-Based Parser for Universal Dependency Parsing

AntNLP at CoNLL 2018 Shared Task: A Graph-Based Parser for Universal Dependency Parsing
复制标题

DOI:
10.18653/v1/k18-2025
复制
发表时间:
2018
期刊:
--
影响因子:
--
通讯作者:
Tao Ji;Yufang Liu;Yijun Wang-;Yuanbin Wu;Man Lan
Tao Ji;Yufang Liu;Yijun Wang-;Yuanbin Wu;Man Lan
中科院分区:
其他
文献类型:
--
作者:
Tao Ji;Yufang Liu;Yijun Wang-;Yuanbin Wu;Man Lan

文献摘要

被引文献

相似文献

我们描述了我们提交给CoNLL 2018 UD共享任务的系统(AntNLP)中基于图的依赖解析器。我们使用双向lstm来获得单词表示,然后使用双仿射指针网络来计算候选依赖边的分数,然后使用MST算法来获得最终的依赖树。从官方测试结果来看,我们的系统获得了70.90 LAS F1分数(排名9/26),55.92 MLAS(排名10/26)和60.91 BLEX(排名8/26)。
We describe the graph-based dependency parser in our system (AntNLP) submitted to the CoNLL 2018 UD Shared Task. We use bidirectional lstm to get the word representation, then a bi-affine pointer networks to compute scores of candidate dependency edges and the MST algorithm to get the final dependency tree. From the official testing results, our system gets 70.90 LAS F1 score (rank 9/26), 55.92 MLAS (10/26) and 60.91 BLEX (8/26).