Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Semantic Communication Enhanced by Knowledge Graph Representation Learning

Nour Hello, Paolo Di Lorenzo, Emilio Calvanese Strinati

TL;DR

This work tackles semantic communications by representing knowledge as graphs and transmitting their meaning through compact semantic embeddings. It introduces an end-to-end framework that cascades large language models with graph neural networks to produce node and relation embeddings, enabling two encoding variants: one that leverages graph topology (Enc_llm_gnn) and one that Compresses node features independently (Enc_llm_ffn). A mutual-information–driven channel coding scheme and dual objective losses train the system to recover accurate knowledge graphs at the receiver, with node and relation decoding performed by a Node Classifier and a Transformer-based Relation Classifier. Numerical results on WebNLG demonstrate substantial compression gains (up to about 24x) and robustness in AWGN channels, with the graph-aware encoder consistently outperforming the baselines, especially in low-SNR regimes. This approach advances practical knowledge-centric semantic transmission for multi-agent AI ecosystems, enabling efficient, meaningful exchanges of structured knowledge across wireless links.

Abstract

This paper investigates the advantages of representing and processing semantic knowledge extracted into graphs within the emerging paradigm of semantic communications. The proposed approach leverages semantic and pragmatic aspects, incorporating recent advances on large language models (LLMs) to achieve compact representations of knowledge to be processed and exchanged between intelligent agents. This is accomplished by using the cascade of LLMs and graph neural networks (GNNs) as semantic encoders, where information to be shared is selected to be meaningful at the receiver. The embedding vectors produced by the proposed semantic encoder represent information in the form of triplets: nodes (semantic concepts entities), edges(relations between concepts), nodes. Thus, semantic information is associated with the representation of relationships among elements in the space of semantic concept abstractions. In this paper, we investigate the potential of achieving high compression rates in communication by incorporating relations that link elements within graph embeddings. We propose sending semantic symbols solely equivalent to node embeddings through the wireless channel and inferring the complete knowledge graph at the receiver. Numerical simulations illustrate the effectiveness of leveraging knowledge graphs to semantically compress and transmit information.

Semantic Communication Enhanced by Knowledge Graph Representation Learning

TL;DR

This work tackles semantic communications by representing knowledge as graphs and transmitting their meaning through compact semantic embeddings. It introduces an end-to-end framework that cascades large language models with graph neural networks to produce node and relation embeddings, enabling two encoding variants: one that leverages graph topology (Enc_llm_gnn) and one that Compresses node features independently (Enc_llm_ffn). A mutual-information–driven channel coding scheme and dual objective losses train the system to recover accurate knowledge graphs at the receiver, with node and relation decoding performed by a Node Classifier and a Transformer-based Relation Classifier. Numerical results on WebNLG demonstrate substantial compression gains (up to about 24x) and robustness in AWGN channels, with the graph-aware encoder consistently outperforming the baselines, especially in low-SNR regimes. This approach advances practical knowledge-centric semantic transmission for multi-agent AI ecosystems, enabling efficient, meaningful exchanges of structured knowledge across wireless links.

Abstract

This paper investigates the advantages of representing and processing semantic knowledge extracted into graphs within the emerging paradigm of semantic communications. The proposed approach leverages semantic and pragmatic aspects, incorporating recent advances on large language models (LLMs) to achieve compact representations of knowledge to be processed and exchanged between intelligent agents. This is accomplished by using the cascade of LLMs and graph neural networks (GNNs) as semantic encoders, where information to be shared is selected to be meaningful at the receiver. The embedding vectors produced by the proposed semantic encoder represent information in the form of triplets: nodes (semantic concepts entities), edges(relations between concepts), nodes. Thus, semantic information is associated with the representation of relationships among elements in the space of semantic concept abstractions. In this paper, we investigate the potential of achieving high compression rates in communication by incorporating relations that link elements within graph embeddings. We propose sending semantic symbols solely equivalent to node embeddings through the wireless channel and inferring the complete knowledge graph at the receiver. Numerical simulations illustrate the effectiveness of leveraging knowledge graphs to semantically compress and transmit information.
Paper Structure (10 sections, 9 equations, 4 figures)

This paper contains 10 sections, 9 equations, 4 figures.

Table of Contents

  1. Introduction
  2. System Model
  3. Semantic Encoders
  4. Channel Coding
  5. Semantic Decoder
  6. Nodes Classifier
  7. Relation Classifier
  8. Knowledge Graph Representation Learning
  9. Numerical Results
  10. Conclusion

Figures (4)

  • Figure 1: System model of semantic representation learning on graphs
  • Figure 2: Node Classification Accuracy vs. Node Embedding Dimensionality.
  • Figure 3: F1 Score of received and source knowledge graphs matching vs. Signal To Noise Ratio when Dealing with AWGN Channel.
  • Figure 4: Transmitted information in bits per graph versus the number of nodes in the graph