China
Africa
Explore chapters and articles related to this topic
Natural Language Processing
Published in Vishal Jain, Akash Tayal, Jaspreet Singh, Arun Solanki, Cognitive Computing Systems, 2021
V. Vishnuprabha, Lino Murali, Daleesha M. Viswanathan
Semantic role labeling is to identify a verb or a predicate and its arguments in a sentence [1]. Each argument identified is labeled based on its semantic relation with the predicate. Semantic role labeling within the sentence makes the meaning representation independent of syntactic arrangement. Both supervised and unsupervised methods are used in semantic role labeling. The semantic roles can come variously from resources such as PropBank, FrameNet, or VerbNet. It can also be extended to similar semantic roles that are introduced by other POS, such as nominal or adjectival elements [2]. Semantic role labeling is also called as thematic role labeling, case role assignment, or even shallow semantic parsing [1].
Semantic Role Labeling Based on Valence Structure and Deep Neural Network
Published in IETE Journal of Research, 2023
The experimental corpus was taken from Chinese PropBank 2.0 and Chinese NomBank 1.0. CTB is a corpus publicly published by the Linguistic Data Consortium (LDC), which provides a common training and testing platform for Chinese syntactic parsing research. PropBank 2.0 is a semantic corpus with Verbal predicates annotated by University of Pennsylvania based on the Penn TreeBank 5.1 syntactic parsing corpus. Chinese NomBank 1.0 was developed to compensate for the limitations of PropBank using only verbs as predicates; it annotates the nominal predicates and their semantic roles in Penn TreeBank 5.1. In order to balance various corpus sources in the training set, development set and test set, referring to the experimental data division of Xue [2], this article take 1296 files of 648 files (chtb_081.fid-chtb 899.fid) in Chinese PropBank2.0 and NomBank1.0 respectively for training set; 80 files of 40 files (chtb_ 041.fid-chtb_080.fid) respectively for development sets; 144 files of 72 files (chtb_001.fid-chtb_040.fid and chtb_900.fid-chtb_931.fid) respectively for test sets. Among them, the number of verbal predicates contained in the training set, development set and test set are 31361, 2060 and 3599 respectively; the number of nominal predicates contained in the training set, development set and test set are 8642,731 and 1124 respectively. SVM classifiers are used in all experiments in this paper, and SVM classifiers use polynomial kernel functions. The parameters of the model are estimated from the training set using maximum likelihood method. Adjustment settings for training parameters are performed on the development set. The performance evaluation of the model and semantic role labeling methods was performed on the test set.