基于熵函数的语义信息边界研究

摘要：经典通信系统需要承载的数据量越来越多，但受到香浓极限约束，其容量可提升空间变得越来越小。人工智能的不断发展助力语义通信不断成熟，语义通信越来越被证明是未来“智能体”之间通信的新范式。然而语义通信的理论基础却不像经典通信那样扎实，语义度量理论发展缓慢且极不成熟。提出了语义度量的4条基本假设，试图将经典信息理论与强语义信息理论等归入统一的语义度量理论中，实现概率视角下语义度量方式与真性距离视角下语义度量方式的有机结合。在梳理语义度量理论发展历程的同时，试图将其中的要素融汇贯通，进一步给出相关概念，并基于熵函数对语义信息的边界进行探索。

关键词：Shannon极限；语义通信；人工智能；语义度量

Abstract: Classical communication systems are increasingly burdened with larger volumes of data. However, the inherent capacity of these systems is limited by the Shannon limit, leaving little room for further enhancement. The ongoing advancement of artificial intelligence is also facilitating the maturation of semantic communication, steadily establishing itself as a new communication paradigm for future "intelligent agents." Despite being proposed concurrently with digital communication, the theoretical foundation of semantic communication is not as robust as that of classical communication, and the development of theories for semantic measurement has been slow and underdeveloped. Four basic assumptions of semantic measurement are proposed, aiming to incorporate classical information theory and strong semantic information theory into a unified semantic measurement theory, and realize the organic combination of semantic measurement methods from the probability perspective and those from the truth distance perspective. This paper attempts to explore the boundary of semantic information based on the entropy function, while combing through the development of semantic measurement theory and trying to integrate its elements.

Keywords: Shannon limit; semantic communication; artificial intelligence; semantic measurement