Better Combine Them Together! Integrating Syntactic Constituency and Dependency Representations for Semantic Role Labeling

1Wuhan University,   2Guangdong University of Foreign Studies


Structural syntax knowledge has been proven effective for semantic role labeling (SRL), while existing works mostly use only one singleton syntax, such as either syntactic dependency or constituency tree. In this project, we explore the integration of heterogeneous syntactic representations for SRL. We first consider a TreeLSTM-based integration, collaboratively learning the phrasal boundaries from the constituency and the semantic relations from dependency. We further introduce a labelaware GCN solution for simultaneously modeling the syntactic edges and labels. Experimental results demonstrate that by effectively combining the heterogeneous syntactic representations, our methods yield task improvements on both span-based and dependencybased SRL. Also our system achieves new state-of-the-art SRL performances, meanwhile bringing explainable task improvements.


▶   Backbone SRL Framework:

▶   TreeLSTM-based HeSyFu:

▶   Label-aware GCN-based HeSyFu:


We aim to answer the following five questions.

• Q1. Whether the combination of constituent and dependency syntax can really improve SRL?

A1. combining two types of syntax information is more helpful than just using either one of them.

• Q2. If yes, how much will such improvements be for the dependency- and span-based SRL?

A2. the improvement for span-based SRL is more obvious than dependency-based one.

• Q3. How different will the results be by employing the TreeLSTM or GCN encoder?

A3. GCN performs better than TreeLSTM.

• Q4. Can SRL be further improved by leveraging syntactic labels?

A4. syntactic labels are quite helpful for SRL.

• Q5. What kind of associations can be discovered between SRL structures and these heterogeneous syntactic structures?

A5. SRL and both kinds of syntactic structures have strong associations and should be exploited for mutual benefits.



    title = "Better Combine Them Together! Integrating Syntactic Constituency and Dependency Representations for Semantic Role Labeling",
    author = "Fei, Hao  and
      Wu, Shengqiong  and
      Ren, Yafeng  and
      Li, Fei  and
      Ji, Donghong",
    booktitle = "Findings of the Association for Computational Linguistics: ACL-IJCNLP 2021",
    year = "2021",
    publisher = "Association for Computational Linguistics",
    pages = "549--559",