{"ID":2874103,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2509.04801","arxiv_id":"2509.04801","title":"KGRAG-SC: Knowledge Graph RAG-Assisted Semantic Communication","abstract":"The state-of-the-art semantic communication (SC) schemes typically rely on end-to-end deep learning frameworks that lack interpretability and struggle with robust semantic selection and reconstruction under noisy conditions. To address this issue, this paper presents KGRAG-SC, a knowledge graph-assisted SC framework that leverages retrieval-augmented generation principles. KGRAG-SC employs a multi-dimensional knowledge graph, enabling efficient semantic extraction through community-guided entity linking and GraphRAG-assisted processing. The transmitter constructs minimal connected subgraphs that capture essential semantic relationships and transmits only compact entity indices rather than full text or semantic triples. An importance-aware adaptive transmission strategy provides unequal error protection based on structural centrality metrics, prioritizing critical semantic elements under adverse channel conditions. At the receiver, large language models perform knowledge-driven text reconstruction using the shared knowledge graph as structured context, ensuring robust semantic recovery even with partial information loss. Experimental results demonstrate that KGRAG-SC achieves superior semantic fidelity in low Signal-to-Noise Ratio (SNR) conditions while significantly reducing transmission overhead compared to traditional communication methods, highlighting the effectiveness of integrating structured knowledge representation with generative language models for SC systems.","short_abstract":"The state-of-the-art semantic communication (SC) schemes typically rely on end-to-end deep learning frameworks that lack interpretability and struggle with robust semantic selection and reconstruction under noisy conditions. To address this issue, this paper presents KGRAG-SC, a knowledge graph-assisted SC framework th...","url_abs":"https://arxiv.org/abs/2509.04801","url_pdf":"https://arxiv.org/pdf/2509.04801v1","authors":"[\"Dayu Fan\",\"Rui Meng\",\"Song Gao\",\"Xiaodong Xu\"]","published":"2025-09-05T04:33:12Z","proceeding":"eess.SP","tasks":"[\"eess.SP\"]","methods":"[\"RAG\",\"Language Model\"]","has_code":false}