Let us first agree on what the term "semantics" means: An unorthodox approach to an age-old debate
Traditionally, semantics has been seen as a feature of human language. The advent of the information era has led to its widespread redefinition as an information feature. Contrary to this praxis, I define semantics as a special kind of information. Revitalizing the ideas of Bar-Hillel and Carnap I have recreated and re-established the notion of semantics as the notion of Semantic Information. I have proposed a new definition of information (as a description, a linguistic text, a piece of a story or a tale) and a clear segregation between two different types of information - physical and semantic information. I hope, I have clearly explained the (usually obscured and mysterious) interrelations between data and physical information as well as the relation between physical information and semantic information. Consequently, usually indefinable notions of “information”, “knowledge”, “memory”, “learning” and “semantics” have also received their suitable illumination and explanation.
💡 Research Summary
The paper tackles a long‑standing philosophical debate by redefining “semantics” not as an exclusive property of human language but as a particular kind of information. The author begins by observing that traditional semantics has been treated as a linguistic phenomenon, while the information age has blurred the line between language and data, leading many to conflate “information” with “meaning.” To restore a clear conceptual distinction, the author proposes a two‑tiered model of information: physical information and semantic information.
Physical information is identified with raw, observable data—sensor readings, voltage levels, temperature measurements, etc.—that can be quantified using classical Shannon‑type entropy measures. It is objective, context‑independent, and directly tied to the physical world. Semantic information, by contrast, emerges only when an observer interprets physical data within a specific context, applying background knowledge, cultural conventions, or task‑related goals. The same set of physical measurements can give rise to entirely different semantic contents for different agents or in different situations.
To ground this distinction, the author revisits the classic works of Bar‑Hillel and Carnap, arguing that semantic information should not be seen as a subset of physical information but as an orthogonal layer that “sits on top of” the physical data. In this view, meaning is not reducible to raw signal patterns; rather, it provides the structural scaffolding that selects, groups, and re‑assembles physical symbols into higher‑order narratives.
The paper then formalizes the transformation pipeline: (1) Data—raw observations; (2) Physical Information—statistically processed, compressed representations of the data; (3) Semantic Information—the result of cognitive operations such as pattern recognition, concept mapping, and schema activation applied to the physical information. Within this pipeline, “memory” is modeled as a repository of previously generated semantic information, while “learning” is the mechanism that updates the mapping between physical and semantic layers in response to new inputs.
By redefining “information” as a descriptive, linguistic text or story, the author creates a unified terminology that links several traditionally vague concepts. “Knowledge” becomes a structured collection of semantic information; “information” itself is the combination of physical and semantic components; “learning” is the continual refinement of semantic structures; and “semantics” is the property of information that enables it to convey meaning. This taxonomy resolves many ambiguities that have plagued interdisciplinary discussions among linguists, philosophers, and information scientists.
The implications of this framework are far‑reaching. In artificial intelligence, it suggests a clear separation between low‑level signal processing (the domain of deep neural networks) and high‑level meaning construction (the domain of symbolic reasoning or knowledge graphs). In cognitive science, it offers a formal model for how the brain might store raw sensory traces while simultaneously building narrative representations that guide behavior. In data science, it provides a rationale for why purely statistical models often fail to capture the “why” behind patterns, emphasizing the need for contextual, domain‑specific knowledge bases.
The conclusion calls for future work on quantifying semantic information, measuring the efficiency of transformations between physical and semantic layers, and designing mechanisms for shared meaning between humans and machines. By re‑establishing semantics as a distinct, information‑based construct, the paper lays groundwork for a more coherent theory of meaning that can be operationalized across disciplines.