Large-scale networks of real-world entities and their relations.
A Knowledge Graph (KG) is a large-scale semantic network that represents a collection of interlinked descriptions of entities—objects, events, or concepts—where entities are nodes and the relationships between them are edges. KGs store information in a structured, graph-based format, typically as a set of 'triples,' each consisting of a subject, a predicate, and an object (e.g., 'Socrates - is a - Philosopher'). This structure allows for the integration of data from diverse sources and enables machines to understand the context and relationships within the data, moving beyond simple keyword matching. For instance, Google's Knowledge Graph powers its infoboxes in search results. When you search for a famous person, it doesn't just find web pages with their name; it understands that the person is an entity with properties like a birth date, profession, and relationships to other entities like family members or works they created. This rich, connected data allows the search engine to answer complex queries directly, like 'What movies has Tom Hanks' wife been in?'. Building a KG involves two main steps: knowledge extraction and knowledge fusion. Extraction involves pulling structured information from various sources like text, databases, and web pages. Fusion involves cleaning and integrating this information, resolving ambiguities (e.g., distinguishing between 'Apple' the company and 'apple' the fruit) and linking entities together. KGs are a powerful bridge between symbolic AI and data-driven methods, providing structured knowledge that can be used to enhance machine learning models, improve recommendation systems, and build more intelligent question-answering systems.