【摘要】 从语义信息论的角度出发，探讨语义知识库的压缩原理与表征极限。考虑用语义率失真函数对语义知识库的压缩性能与表达能力进行量化建模，并基于语义信息论中的同义映射关系，考虑收发端语义知识库完全匹配和不匹配的场景。进一步引入了语义失真函数以及语义知识库匹配度，量化分析语义知识库一致性的影响。在CUB数据集上的仿真结果表明，随着语义知识库规模的增加，语义通信系统的压缩效率显著提升。同时，提升收发端语义知识库的匹配度对于实现高效语义通信至关重要，从而验证了所提出理论框架与算法模型的有效性。更多还原

【Abstract】 This paper investigates the fundamental limits of knowledge base representation and compression from the perspective of semantic information theory.A semantic rate-distortion function is introduced to quantify the tradeoff between compression efficiency and representational fidelity.Building on synonymic mapping relations in semantic information theory,we consider both fully aligned and mismatched semantic knowledge bases at the transmitter and receiver.A semantic distortion function and a knowledge base alignment metric are further defined to model and evaluate the impact of semantic consistency.Simulation results on the CUB dataset demonstrate that larger knowledge bases enable more efficient semantic compression.Moreover,a higher degree of alignment between transmitter and receiver knowledge bases significantly enhances semantic communication performance,thereby validating the effectiveness of the proposed theoretical framework and algorithmic design.更多还原