An Easier and Efficient Framework to Annotate Semantic Roles: Evidence from the Chinese AMR Corpus

Abstract

Semantic role labeling (SRL) is a fundamental task in Chinese language processing, but there are three major problems about the construction of SRL corpora. First, disagreements occurred in previous studies over the definition and number of semantic roles. Second, it is hard for static predicate frames to cover dynamic predicate usages. Third, it is unable to annotate the dropped semantic roles. Abstract Meaning Representation (AMR) is a new method which provides a better solution to the above problems. The researchers use 5,000 sentences in the Chinese AMR corpus to make a comparison between AMR and other SRL resources. Data analysis shows that within the framework of AMR, it is easier to annotate semantic roles based on simplified distinction between core and non-core roles. In addition, 1,045 tokens of dropped roles are annotated under this new framework. This study indicates that AMR offers a better solution for Chinese SRL and sentence meaning processing.