CISum: Learning Cross-modality Interaction to Enhance Multimodal Semantic Coverage for Multimodal Summarization
This work addresses the challenge of effectively summarizing multimodal content for applications like news or social media, though it appears incremental by building on prior methods focused on textual metrics.
The paper tackles the problem of multimodal summarization by proposing CISum, a framework that learns cross-modality interactions to enhance semantic coverage across text and images, resulting in improved performance on multimodal metrics while maintaining strong ROUGE and BLEU scores.
Multimodal summarization (MS) aims to generate a summary from multimodal input. Previous works mainly focus on textual semantic coverage metrics such as ROUGE, which considers the visual content as supplemental data. Therefore, the summary is ineffective to cover the semantics of different modalities. This paper proposes a multi-task cross-modality learning framework (CISum) to improve multimodal semantic coverage by learning the cross-modality interaction in the multimodal article. To obtain the visual semantics, we translate images into visual descriptions based on the correlation with text content. Then, the visual description and text content are fused to generate the textual summary to capture the semantics of the multimodal content, and the most relevant image is selected as the visual summary. Furthermore, we design an automatic multimodal semantics coverage metric to evaluate the performance. Experimental results show that CISum outperforms baselines in multimodal semantics coverage metrics while maintaining the excellent performance of ROUGE and BLEU.