Recognizing Functional Pragmatics in Chinese Discourses on Enhancing Paragraph Representation and Deep Differential Amplifier.

Discourse functional pragmatics recognition focuses on identifying the functions of discourse paragraphs, which is a significant research direction in natural language processing. To obtain better paragraph representation and alleviate the issue of imbalanced data distribution, we propose a Chinese discourse Functional Pragmatics Recognition model (FPR) based on enhancing paragraph representation and deep differential amplifier. More specifically, we first combine two different encoding methods to obtain paragraph encoding which contains much richer word-level information. And then we apply a deep differential amplifier to amplify the difference between paragraphs, which consists of residual structure and iterative structures. Additionally, we give more attention to the minority classes by adjusting the weights of the minority and majority classes in the objected function. Experimental results on MCDTB 2.0 show that our model FPR outperforms the state-of-the-art models.