YAGO

YAGO is a large, semantically structured knowledge base that consolidates information about entities and their relationships into a single, high-precision ontology and graph. It represents a wide range of entities, including persons, places, organizations, and other things, together with attributes and inter-entity relations such as isA, bornIn, locatedIn, and relatedTo. The project emphasizes a typed, multi-relational approach to enable robust reasoning and querying.

The data model in YAGO combines entities, classes, and relations in a graph structure. It uses a rigorous type system and disjointness constraints to improve accuracy and reduce ambiguity. Information is stored with explicit types and provenance, supporting reliable inference and cross-source validation. The project aims for high precision, often trading some completeness for the reliability of inferred facts.

Origins and sources: YAGO was developed by researchers including Fabian Suchanek, Gjergji Kasneci, and Gerhard Weikum. It integrates data from Wikipedia, WordNet, and GeoNames, aligning entities and relations across these sources to form a coherent knowledge base. The extraction process blends automatic information gathering with quality-control rules and manual checks to resolve ambiguities and remove conflicting data. The resulting dataset is designed to be readily reusable by other systems and researchers.

Access and applications: YAGO is released in standard formats such as RDF, Turtle, and N-Triples, enabling integration with other knowledge bases and tools. It has been used for research in link prediction, entity disambiguation, question answering, and ontology alignment, and has served as a benchmark for data integration and knowledge-graph quality studies.

Related knowledge bases include DBpedia, Freebase, and Wikidata, with YAGO influencing approaches to structured data extraction and validation.