Foundations: From Data to Knowledge

The transformation of raw observations into meaningful understanding represents one of the most profound capabilities of human cognition. When we encounter a scattered collection of numbers—847, 1024, 23.5, 6.28—we perceive mere symbols without context. Yet when these same values emerge within a specific framework, such as a recipe scaled for a dinner party or coordinates defining a location, they acquire significance. This cognitive alchemy, this passage from inert data to actionable knowledge, sits at the heart of what this book calls Semantic Cartography: the discipline devoted to architecting structured representations of knowledge that machines can navigate, reason about, and exploit.

Knowledge graphs have emerged as the predominant paradigm for achieving this transformation at industrial and scientific scale. Unlike simpler data structures that store facts in isolation, knowledge graphs capture the intricate web of relationships that bind concepts together, enabling both human users and software systems to traverse paths of meaning and derive inferences that transcend the explicit statements encoded within them. The past two decades have witnessed remarkable advances in knowledge graph technology, from academic research prototypes to mission-critical infrastructure deployed by the world's largest technology companies. Yet despite this proliferation, a coherent understanding of what knowledge graphs truly are, how they evolved from their conceptual predecessors, and what distinguishes them from superficially similar technologies remains elusive to many practitioners.

This opening chapter establishes the conceptual foundations necessary for the rigorous study of knowledge graphs. We begin by examining the evolutionary trajectory from raw data through information to knowledge, establishing the philosophical and practical distinctions that inform subsequent technical discussions. This journey through the data-information-knowledge-wisdom hierarchy provides essential context for understanding why knowledge graphs were developed and what problems they were designed to solve. We then trace the historical development of knowledge representation systems, from the early expert systems of the 1970s through the frame-based representations that dominated the 1980s to the modern graph-structured knowledge bases that underpin contemporary applications. This historical perspective reveals both the intellectual lineages connecting current approaches to their predecessors and the lessons learned from decades of research and experimentation.

Central to our investigation is the establishment of precise terminology. The fields of knowledge representation, semantic technologies, and graph databases have generated considerable conceptual confusion through inconsistent usage of terms such as ontology, schema, knowledge base, and knowledge graph. This chapter provides rigorous definitions and distinguishing characteristics for these fundamental concepts, enabling precise communication throughout the remainder of the book. We examine not only the technical dimensions of these distinctions but also the philosophical underpinnings that inform different theoretical positions.

The philosophical dimensions of knowledge representation deserve particular attention because they illuminate fundamental challenges that no technical approach can fully resolve. The frame problem, first articulated by John McCarthy and Patrick Hayes in 1969, exposes the difficulty of specifying which facts remain relevant as an intelligent agent considers alternative courses of action. The symbol grounding problem, formulated by Stevan Harnad in 1990, questions how the arbitrary symbols employed by computational systems acquire meaning through reference to the external world. The representation trap describes the tendency to confuse the map for the territory, creating formal structures that appear meaningful but fail to capture genuine understanding. Awareness of these philosophical issues does not paralyze technical work but rather guides the development of approaches that acknowledge their limitations while providing practical utility.

The W3C standards ecosystem—comprising the Resource Description Framework (RDF), the Web Ontology Language (OWL), and the SPARQL Protocol and RDF Query Language—provides the technological foundation for contemporary knowledge graph implementations. Understanding these standards in historical context reveals both their design principles and their intended scope, enabling practitioners to employ them effectively while recognizing their boundaries. We examine not only the technical specifications themselves but also the design decisions and trade-offs that shaped their development.

Architectural paradigms for knowledge graphs vary significantly across different deployment contexts. The Linked Open Data cloud demonstrated the possibility of creating a global knowledge graph through distributed publication of structured data following common standards. Google's Knowledge Vault illustrated how probabilistic methods could address the challenges of knowledge acquisition at web scale. Enterprise knowledge graphs address organizational requirements for data integration, governance, and operational reliability that differ substantially from research or consumer applications. Comparative analysis of these paradigms illuminates the design space and informs architectural decisions for new implementations.

Finally, we address the question that closes this chapter and opens subsequent discussions: what truly distinguishes a knowledge graph from a conventional graph database? This question has practical consequences for technology selection, architectural design, and evaluation criteria. We establish not only conceptual distinctions but also operational characteristics that enable practitioners to assess whether a given implementation warrants the designation of knowledge graph and to evaluate its quality and completeness according to principled criteria.

The Data-Information-Knowledge Continuum

Understanding knowledge graphs requires first understanding what knowledge itself entails and how it differs from the raw materials from which it emerges. The data-information-knowledge-wisdom hierarchy, often attributed to various authors including Russell Ackoff and later popularized by Thomas Davenport, provides a foundational framework for this investigation. While the strict hierarchical nature of this model has been questioned by researchers who emphasize more fluid transitions between levels, the framework nonetheless illuminates crucial distinctions that inform the design of knowledge graph architectures.

Data consists of raw, unprocessed facts without context or interpretation. A temperature reading of 98.6 degrees, the number 42, the string "2024-01-15"—these are data points that exist independently of meaning or purpose. Data becomes data through its form: organized symbols that can be stored, transmitted, and processed by computational systems. The value of data lies in its potential, as yet unrealized, to inform decisions or support reasoning when appropriately contextualized.

Information emerges when data is placed within a meaningful context, enabling interpretation and answering questions of the form "who," "what," "where," and "when." The same temperature reading becomes information when we understand it as a patient's body temperature measured at a specific time in a particular clinical setting. Information processing involves structuring data in ways that reveal relationships and support specific analytical purposes. The transformation from data to information is fundamentally about answering questions and reducing uncertainty about specific matters.

Knowledge represents a higher-order understanding that incorporates not only isolated facts but also the relationships between them and the capacity to apply this understanding to action. Knowledge