Hybrid Transformer with Multi-level Fusion for Multimodal Knowledge Graph Completion

SIGIR '22: Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval(2022)

引用 80|浏览424
暂无评分
摘要
Multimodal Knowledge Graphs (MKGs), which organize visual-text factual knowledge, have recently been successfully applied to tasks such as information retrieval, question answering, and recommendation system. Since most MKGs are far from complete, extensive knowledge graph completion studies have been proposed focusing on the multimodal entity, relation extraction and link prediction. However, different tasks and modalities require changes to the model architecture, and not all images/objects are relevant to text input, which hinders the applicability to diverse real-world scenarios. In this paper, we propose a hybrid transformer with multi-level fusion to address those issues. Specifically, we leverage a hybrid transformer architecture with unified input-output for diverse multimodal knowledge graph completion tasks. Moreover, we propose multi-level fusion, which integrates visual and text representation via coarse-grained prefix-guided interaction and fine-grained correlation-aware fusion modules. We conduct extensive experiments to validate that our MKGformer can obtain SOTA performance on four datasets of multimodal link prediction, multimodal RE, and multimodal NER1. https://github.com/zjunlp/MKGformer.
更多
查看译文
关键词
knowledge graph completion, multimodal, relation extraction, named entity recognition
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络