ThreadsGAN: Enhancing Coherence and Diversity in Discussion Thread Generation
Keywords: discussion threads, generative adversarial network, natural language generating
Abstract: Current research on generating discussion threads faces challenges in coherence, interactivity, and multi-topic handling, which are crucial for meaningful responses. This paper introduces threadsGAN, a model that enhances thread generation by incorporating multi-topic and social response intention tags. By leveraging BERT and Transformer, threadsGAN ensures contextual coherence and manages topic consistency. Additionally, it employs conditional generation to align responses with specific discussion contexts, and its CNN-based discriminator assesses response quality by evaluating similarity between generated and real responses, improving overall performance in generating realistic and contextually appropriate discussion threads.
Primary Area: generative models
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 9605
Loading