Overview of Argumentative Text Understanding for AI Debater Challenge

Yuan, Jian; Cheng, Liying; He, Ruidan; Li, Yinzi; Wei, Zhongyu; Liu, Qin; Shen, Chenhui; Zhang, Shuonan; Sun, Chengjie; Si, Luo; Jiang, Changjian; Huang, Xuanjing

doi:10.1007/978-3-030-88483-3_45

Cited by 1 publication

(1 citation statement)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, in the subsequent model prediction, only 13 reviews were predicted to have the emotion of "fear," which is consistent with the actual data distribution. The debate dataset [44] originates from dozens of famous Chinese-language debate competitions in recent years. The text for each single section and single-party statement of each competition was obtained through speech-to-text conversion and manual verification and was further annotated by annotators for proposition sentences and interactive proposition pairs.…”

Section: Experiments and Results Analysis 41 Datasetmentioning

confidence: 99%

Utilizing Latent Diffusion Model to Accelerate Sampling Speed and Enhance Text Generation Quality

Li,

Zhang,

Zheng

2024

Electronics

View full text Add to dashboard Cite

Diffusion models have achieved tremendous success in modeling continuous data modalities, such as images, audio, and video, yet their application in discrete data domains (e.g., natural language) has been limited. Existing methods primarily represent discrete text in a continuous diffusion space, incurring significant computational overhead during training and resulting in slow sampling speeds. This paper introduces LaDiffuSeq, a latent diffusion-based text generation model incorporating an encoder–decoder structure. Specifically, it first employs a pretrained encoder to map sequences composed of attributes and corresponding text into a low-dimensional latent vector space. Then, without the guidance of a classifier, it performs the diffusion process for the sequence’s corresponding latent space. Finally, a pretrained decoder is used to decode the newly generated latent vectors, producing target texts that are relevant to themes and possess multiple emotional granularities. Compared to the benchmark model, DiffuSeq, this model achieves BERTScore improvements of 0.105 and 0.009 on two public real-world datasets (ChnSentiCorp and a debate dataset), respectively; perplexity falls by 3.333 and 4.562; and it effectively quadruples the text generation sampling speed.

show abstract

Section: Experiments and Results Analysis 41 Datasetmentioning

confidence: 99%