Fundamentals and Applications of Knowledge Graphs

This article explains the basic concepts of knowledge graphs, including their background, definition, and typical applications.

01 Knowledge Graph Background

Before defining a knowledge graph, the article distinguishes between knowledge and graph . Knowledge is described as the systematic understanding derived from human practice, encompassing facts, information, descriptions, and skills. The DIKW hierarchy (Data‑Information‑Knowledge‑Wisdom) is introduced to illustrate how raw data is refined into knowledge and wisdom.

Examples show how isolated data (e.g., "226.1 cm") become information when contextualized (e.g., "Yao Ming's arm span is 226.1 cm") and further become knowledge when abstracted and integrated.

A graph is defined as a structure of nodes (vertices) and edges (links). The article presents a simple graph with six nodes and seven edges and explains that a knowledge graph represents knowledge using such a graph, where nodes denote semantic entities or concepts and edges denote various semantic relations.

The Resource Description Framework (RDF) representation <subject, predicate, object> is introduced as a common way to encode knowledge triples.

02 Knowledge Graph Definition

The article reviews the evolution of the knowledge‑graph concept: from semantic networks (Richens) to the introduction of ontology in the 1980s, the emergence of the Semantic Web (Tim Berners‑Lee, 1998), and the concept of Linked Data (2006). It notes that Google commercialized the idea in 2012 and that knowledge graphs are closely related to AI, NLP, knowledge representation, databases, and the Web.

Several academic definitions are quoted, emphasizing that a knowledge graph is a structured semantic knowledge base composed of entity‑relation‑entity triples, attributes, and a graph‑like network of concepts.

03 Typical Applications of Knowledge Graphs

Medical domain: Platforms like Open PHACTS integrate pharmacological data to support drug discovery; IBM Watson uses knowledge graphs for oncology and other medical tasks.

Financial & investment domain: Companies such as AlphaSense and Kensho build financial knowledge engines that combine knowledge graphs, NLP, and semantic search to analyze market data.

Government & security domain: Palantir’s large‑scale knowledge graphs have been used for intelligence analysis, including counter‑terrorism and fraud investigations.

E‑commerce domain: Alibaba constructs a billion‑scale product knowledge graph to improve search, recommendation, governance, and fraud detection.

Chatbot domain: Voice assistants like Siri, Microsoft Xiaoice, and various Chinese chatbots rely on extensive knowledge graphs for question answering.

References

[1] Rowley, J. The Wisdom Hierarchy: Representations of the DIKW Hierarchy. Journal of Information and Communication Science, 2007.

[2] Zeleny, M. Management Support Systems: Towards Integrated Knowledge Management. Human Systems Management, 1987.

[3] J. J. Sylvester. On an Application of the New Atomic Theory to the Graphical Representation of the Invariants and Covariants of Binary Quantics. American Journal of Mathematics, 1878.

[4] Berners‑Lee, T. Information Management: A Proposal. CERN‑DD‑89‑001‑OC, 1989.

[5] Liu Q. et al. Overview of Knowledge Graph Construction Techniques. Computer Research and Development, 2016.

[6] Li J. & Hou L. Survey of Knowledge Graph Research. Shanxi University Journal (Natural Science), 2017.

[7] Qi G. et al. Progress in Knowledge Graph Research. Information Engineering, 2017.