Frame-semantic Parsing (FSP) is a challenging and critical task in Natural Language Processing (NLP). Most of the existing studies decompose the FSP task into frame identification (FI) and frame semantic role labeling (FSRL) subtasks, and adopt a pipeline model architecture that clearly causes error propagation problem. However, recent jointly learning models aim to address the above problem and generally treat FSP as a span-level structured prediction task, which, unfortunately, leads to cascading error propagation problem between roles and less-efficient solutions due to huge search space of roles. To address these problems, we reformulate the FSRL task into a target-aware relation classification task and propose a novel and lightweight jointly learning framework that simultaneously processes three subtasks of FSP, including frame identification, argument identification, and role classification. The novel task formulation and jointly learning with interaction mechanisms among subtasks can help improve the overall system performance and reduce the search space and time complexity, compared with existing methods. Extensive experimental results demonstrate that our proposed model significantly outperforms 10 state-of-the-art models in terms of F1 score across two benchmark datasets.
Frame-semantic parsing (FSP) aims to extract frame-semantic structures from text. The taskinvolves three subtasks: target identification (TI), frame identification (FI) and frame-semanticrole labeling (FSRL). Most of the previous studies predominantly focus one or two of the threesubtasks, and limited studies focus on constructing a unified framework to systematically solvethe whole FSP task. In addition, how to efficiently utilize lexical units knowledge and improvethe span representations for FSP is still an open question. Therefore, this work presents a novelunified framework for FSP based on marker attention, in which special markers are strategically inserted into the input sequence of the encoder to incorporate the lexical units knowledgeand frame information, as well as to enhance the span representation through attention mechanisms, without using any additional complicated module. Experiments on two public benchmarks demonstrate the effectiveness of our proposed framework.