Multiple POS Dependency-Aware Mixture of Experts for Frame Identification
Frame identification, which is finding the exact evoked frame for a target word in a given sentence, is a fundamental and crucial prerequisite for frame semantic parsing. It is generally seen as a classification task for target words, whose contextual representations are usually obtained using a neu...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2023-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10061199/ |
| _version_ | 1797866975121113088 |
|---|---|
| author | Zhichao Yan Xuefeng Su Qinghua Chai Xiaoqi Han Yunxiao Zhao Ru Li |
| author_facet | Zhichao Yan Xuefeng Su Qinghua Chai Xiaoqi Han Yunxiao Zhao Ru Li |
| author_sort | Zhichao Yan |
| collection | DOAJ |
| description | Frame identification, which is finding the exact evoked frame for a target word in a given sentence, is a fundamental and crucial prerequisite for frame semantic parsing. It is generally seen as a classification task for target words, whose contextual representations are usually obtained using a neural network like BERT as an encoder, and enriched with a joint learning model or the knowledge of FrameNet. However, the distinction at a fine-grained level, such as the delicate differences in the information of syntax and PropBank roles caused by different parts-of-speech (POS) of target words, is neglected. We propose a Multiple POS Dependency-aware Mixture of Experts(MPDaMoE) network that integrates five types of information, consisting of the syntactic information of target words whose POS are nominal, adjectival, adverbial, or prepositional, and the PropBank role information of target words whose POS are only verbal. To better learn such information, a Mixture of Experts network is employed, in which every expert is a Graph Convolutional Network, to incorporate the different dependency information of target words. Our model outperforms state-of-the-art models in experiments on two benchmark datasets, which shows its effectiveness. |
| first_indexed | 2024-04-09T23:32:51Z |
| format | Article |
| id | doaj.art-5d1ef4df77d744c7884db47b1fdc305e |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-04-09T23:32:51Z |
| publishDate | 2023-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-5d1ef4df77d744c7884db47b1fdc305e2023-03-20T23:00:45ZengIEEEIEEE Access2169-35362023-01-0111256042561510.1109/ACCESS.2023.325312810061199Multiple POS Dependency-Aware Mixture of Experts for Frame IdentificationZhichao Yan0https://orcid.org/0000-0002-3336-8886Xuefeng Su1Qinghua Chai2Xiaoqi Han3https://orcid.org/0000-0002-8827-8474Yunxiao Zhao4Ru Li5School of Computer and Information Technology, Shanxi University, Taiyuan, ChinaSchool of Computer and Information Technology, Shanxi University, Taiyuan, ChinaSchool of Foreign Language, Shanxi University, Taiyuan, ChinaSchool of Computer and Information Technology, Shanxi University, Taiyuan, ChinaSchool of Computer and Information Technology, Shanxi University, Taiyuan, ChinaSchool of Computer and Information Technology, Shanxi University, Taiyuan, ChinaFrame identification, which is finding the exact evoked frame for a target word in a given sentence, is a fundamental and crucial prerequisite for frame semantic parsing. It is generally seen as a classification task for target words, whose contextual representations are usually obtained using a neural network like BERT as an encoder, and enriched with a joint learning model or the knowledge of FrameNet. However, the distinction at a fine-grained level, such as the delicate differences in the information of syntax and PropBank roles caused by different parts-of-speech (POS) of target words, is neglected. We propose a Multiple POS Dependency-aware Mixture of Experts(MPDaMoE) network that integrates five types of information, consisting of the syntactic information of target words whose POS are nominal, adjectival, adverbial, or prepositional, and the PropBank role information of target words whose POS are only verbal. To better learn such information, a Mixture of Experts network is employed, in which every expert is a Graph Convolutional Network, to incorporate the different dependency information of target words. Our model outperforms state-of-the-art models in experiments on two benchmark datasets, which shows its effectiveness.https://ieeexplore.ieee.org/document/10061199/Frame identificationsemantic rolessyntactic informationBERTmixture of expertsgraph convolutional network |
| spellingShingle | Zhichao Yan Xuefeng Su Qinghua Chai Xiaoqi Han Yunxiao Zhao Ru Li Multiple POS Dependency-Aware Mixture of Experts for Frame Identification IEEE Access Frame identification semantic roles syntactic information BERT mixture of experts graph convolutional network |
| title | Multiple POS Dependency-Aware Mixture of Experts for Frame Identification |
| title_full | Multiple POS Dependency-Aware Mixture of Experts for Frame Identification |
| title_fullStr | Multiple POS Dependency-Aware Mixture of Experts for Frame Identification |
| title_full_unstemmed | Multiple POS Dependency-Aware Mixture of Experts for Frame Identification |
| title_short | Multiple POS Dependency-Aware Mixture of Experts for Frame Identification |
| title_sort | multiple pos dependency aware mixture of experts for frame identification |
| topic | Frame identification semantic roles syntactic information BERT mixture of experts graph convolutional network |
| url | https://ieeexplore.ieee.org/document/10061199/ |
| work_keys_str_mv | AT zhichaoyan multipleposdependencyawaremixtureofexpertsforframeidentification AT xuefengsu multipleposdependencyawaremixtureofexpertsforframeidentification AT qinghuachai multipleposdependencyawaremixtureofexpertsforframeidentification AT xiaoqihan multipleposdependencyawaremixtureofexpertsforframeidentification AT yunxiaozhao multipleposdependencyawaremixtureofexpertsforframeidentification AT ruli multipleposdependencyawaremixtureofexpertsforframeidentification |