Current methods often overlook the impact of context-dependent sequential text with graph neural networks, leading to an inadequate semantic representation of the text. In this paper, we propose a ...