GLOSSARY
Knowledge graph
A knowledge graph represents entities such as companies, people or products and their named relationships, enabling connected queries across otherwise separate records.
Relationships can change the question worth asking
In B2B sales, companies are connected through ownership, locations, supply relationships or shared decision processes. A knowledge graph can represent these relationships explicitly. It can support questions about which subsidiary belongs to a group or which documented products are used at a site. The potential value comes from connecting existing information. If two similarly named companies are confused, however, the same error can spread through several queries. Correct identification matters as much as the number of stored connections.
Every connection needs meaning and provenance
“Belongs to”, “supplies” and “mentioned” describe different relationships. A graph should preserve their meaning, time validity and supporting information. Evidence helps prevent inferred links from appearing as confirmed facts. Embeddings serve a different purpose, such as finding similar texts; they do not substitute for an explicitly supported relationship. Before building a large dataset, test a bounded set of relevant customer structures. Check identities, outdated relationships and contradictory sources. Include maintenance and correction effort in the value assessment so an impressive visualisation is not mistaken for dependable knowledge.