PulseExploreJournal ClubResearchersJournals
Instagram
HomeJournal ClubExplore
Synapse
⌘+K
Synapse
March 25, 2026IEICE Transactions on Information and SystemsOpen Access

Multimodal Named Entity Recognition with Prior Knowledge from Multimodal Large Models and Text-Directed Fusion

View Full Paper
Ask AI
Bookmark
Share

Authors

LHLi HECXChuang XIONGJDJianyong DUAN

Discussion

Loading...

Member takes

Overview

Multimodal named entity recognition improves prediction using prior knowledge and text-guided integration, suggesting enhanced accuracy in entity extraction.

Key Points

  • The aim is to enhance Multimodal Named Entity Recognition by integrating external knowledge and improving cross-modal fusion.
  • Developed a framework called PKTF with two main stages: prior assisted knowledge generation and entity recognition.
  • Utilized Intern VL2-8B to generate contextual prior knowledge for original text.
  • Designed Text-Max-Directed Fusion Module to focus attention scores based on text guidance.
  • Achieved F1-scores of 75.43% on the Twitter-2015 dataset and 88.74% on the Twitter-2017 dataset.
  • Demonstrated competitive performance compared to existing multimodal models.

Cite This Study

HE et al. (2026) studied this question.

synapsesocial.com/papers/69c37aa8b34aaaeb1a67c7f4https://doi.org/10.1587/transinf.2025kbp0005
View Full Paper
Ask AI
Bookmark
Share