一作SIGIR 2026Poster· Information Retrieval (Short)
MSW-NTM: A Spherical Wasserstein Autoencoder for Multimodal Neural Topic Modeling with LLM-Guided Topic Refinement
Dayu Guo · Zhiwen Luo · Nizar Bouguila · Wentao Fan
摘要
Multimodal topic modeling aims to uncover interpretable semantic structures from documents containing both text and images, while also supporting downstream tasks such as multimodal retrieval. However, existing multimodal neural topic models, typically based on standard Variational Autoencoders (...