Knowledge Fusion via Embeddings from Text, Knowledge Graphs, and Images

TL;DR

This paper introduces a baseline method for integrating knowledge from text, knowledge graphs, and images into unified embeddings, demonstrating the potential of cross-modal knowledge fusion.

Contribution

It evaluates basic fusion methods on existing embeddings to show how combining modalities enhances concept representations.

Findings

Fusion improves concept representation quality

Basic methods show potential for cross-modal knowledge integration

Baseline results establish a foundation for future research

Abstract

We present a baseline approach for cross-modal knowledge fusion. Different basic fusion methods are evaluated on existing embedding approaches to show the potential of joining knowledge about certain concepts across modalities in a fused concept representation.