RADD: Retrieval-Augmented Discrete Diffusion for Multi-Modal Knowledge Graph Completion

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Niu, Guanglin, Li, Bo
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908998994755584
author Niu, Guanglin
Li, Bo
author_facet Niu, Guanglin
Li, Bo
contents Most multi-modal knowledge graph completion (MMKGC) models use one embedding scorer to do both retrieval over the full entity set and final decision making. We argue that this coupling is a core bottleneck: global high-recall search and local fine-grained disambiguation require different inductive biases. Therefore, we propose a Retrieval-Augmented Discrete Diffusion (RADD) framework to decouple retrieve and reranking for MMKGC. A relation-aware multimodal KGE retriever serves as both global retriever and distillation teacher, while a conditional discrete denoiser performs shortlist-level entity-identity generation for reranking. Training combines KGE supervision, denoising cross-entropy, and temperature-scaled distillation from the retriever to the denoiser. At inference, the designed Diff-Rerank first forms a top-$K$ shortlist with the retriever and then reranks it with the denoiser, ensuring that recall is a strict prerequisite for precision. Experiments on three MMKGC benchmarks show that RADD achieves the best performance and consistent gains over strong unimodal, multimodal, and LLM-based baselines, while ablations further verify the contribution of each component.
format Preprint
id arxiv_https___arxiv_org_abs_2604_25693
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle RADD: Retrieval-Augmented Discrete Diffusion for Multi-Modal Knowledge Graph Completion
Niu, Guanglin
Li, Bo
Artificial Intelligence
I.2.7
Most multi-modal knowledge graph completion (MMKGC) models use one embedding scorer to do both retrieval over the full entity set and final decision making. We argue that this coupling is a core bottleneck: global high-recall search and local fine-grained disambiguation require different inductive biases. Therefore, we propose a Retrieval-Augmented Discrete Diffusion (RADD) framework to decouple retrieve and reranking for MMKGC. A relation-aware multimodal KGE retriever serves as both global retriever and distillation teacher, while a conditional discrete denoiser performs shortlist-level entity-identity generation for reranking. Training combines KGE supervision, denoising cross-entropy, and temperature-scaled distillation from the retriever to the denoiser. At inference, the designed Diff-Rerank first forms a top-$K$ shortlist with the retriever and then reranks it with the denoiser, ensuring that recall is a strict prerequisite for precision. Experiments on three MMKGC benchmarks show that RADD achieves the best performance and consistent gains over strong unimodal, multimodal, and LLM-based baselines, while ablations further verify the contribution of each component.
title RADD: Retrieval-Augmented Discrete Diffusion for Multi-Modal Knowledge Graph Completion
topic Artificial Intelligence
I.2.7
url https://arxiv.org/abs/2604.25693