苏大李正华老师受邀对COLING2022 Best Paper做英文报告

每天给你送来NLP技术干货！

---

来自：SudaNLP

受新加坡科技与设计大学（SUTD）陆巍老师、新加坡国立大学（NUS）秦彦霞博士、南京大学黄书剑老师邀请，李正华老师于2022年11月11日上午，对我们的COLING-2022论文进行了长达一小时的英文介绍。

---

视频链接

bilibili：

www.bilibili.com/video/BV1Yd4y1F7ML

Youtube：

www.youtube.com/watch?v=q0XeQpwL7VY

---

汇报摘要

In this talk, I would like to introduce this COLING-2022 paper. Besides the details of the work itself, I also want to discuss the stories behind the work. For example, how did the idea come out; how we developed the idea gradually; how we responded to the negative comments in the previous two submission rounds; etc. I also want to discuss with the audience how our work may be improved and look forward to new ideas.

---

论文摘要

This paper proposes to cast end-to-end span-based SRL as a word-based graph parsing task. The major challenge is how to represent spans at the word level. Borrowing ideas from research on Chinese word segmentation and named entity recognition, we propose and compare four different schemata of graph representation, i.e., BES, BE, BIES, and BII, among which we find that the BES schema performs the best. We further gain interesting insights through detailed analysis.  Moreover, we propose a simple constrained Viterbi procedure to ensure the legality of the output graph according to the constraints of the SRL structure. We conduct experiments on two widely used benchmark datasets, i.e., CoNLL05 and CoNLL12. Results show that our word-based graph parsing approach achieves consistently better performance than previous results, under all settings of end-to-end and predicate-given, without and with pre-trained language models (PLMs). More importantly, our model can parse 669/252 sentences per second, without and with PLMs respectively.

---

李正华老师简介

Zhenghua Li is a full professor at Soochow University Suzhou, China. Zhenghua received the Bachelor, Master, and PhD degrees in computer science from Harbin Institute of Technology, Harbin, China, in 2006, 2008, and 2013, respectively. He joined Soochow University afterward. So far, he has published 40+ top-tier conference papers in the NLP/AI fields, including 9 ACL long papers, as the first author or the corresponding author. His NLPCC-2020 paper had won the best paper award; and his COLING-2022 paper had won the best long paper award. Besides writing papers, he has been actively participating in shared tasks and competitions and has won the first place for multiple times, including tasks on syntax parsing (CoNLL-2009), semantic parsing (SemEval-2019 UCCA, CoNLL-2019 EDS), and grammatical error correction (GEC) (CTC-2021, CGED-2021, WAIC-2022).  Meanwhile, he has been highly interested and constantly engaged in constructing high-quality datasets for research on syntactic parsing (CODT), semantic parsing (MuCPAD), text-to-SQL parsing (DuSQL, SeSQL), and GEC (MuGEC).  He, as PI, has been granted three NSFC projects, and has been closely collaborating with Alibaba, Huawei, and Baidu during the past three years. As a graduate student supervisor, he has 3 students obtaining their PhD degree and about 15 students obtaining their Master degree. Two of them teach and do research in universities, and many work in top-tier IT companies in China. 

本期编辑：周仕林 李正华

---

📝论文解读投稿，让你的文章被更多不同背景、不同方向的人看到，不被石沉大海，或许还能增加不少引用的呦~ 投稿加下面微信备注“投稿”即可。

最近文章

COLING'22 | SelfMix：针对带噪数据集的半监督学习方法

ACMMM 2022 | 首个针对跨语言跨模态检索的噪声鲁棒研究工作

ACM MM 2022 Oral  | PRVR: 新的文本到视频跨模态检索子任务

---

投稿或交流学习，备注：昵称-学校（公司）-方向，进入DL&NLP交流群。

方向有很多：机器学习、深度学习，python，情感分析、意见挖掘、句法分析、机器翻译、人机对话、知识图谱、语音识别等。

记得备注~