Knowledge graph
A knowledge graph is a data model that stores things such as people, customers, projects and documents as connected nodes, so software can follow the relationships between them.
How does a knowledge graph work?
A knowledge graph stores each thing as a node and each relationship as a link between two nodes. A customer can link to a project, the project to a meeting, and the meeting to the people who attended it and the decisions they took. Because the links are stored explicitly, software can answer questions by following them, for example “which customers were discussed in meetings about late projects?”.
How is a knowledge graph different from a database or a data warehouse?
A relational database or a data warehouse stores records in tables, and relationships have to be rebuilt with joins for each question. A knowledge graph treats relationships as first-class data. This makes it a good fit for questions that cross many systems, which is common with organizational knowledge.
Why does a knowledge graph matter for AI assistants?
An AI assistant that answers from one document or one system sees only part of the picture. An assistant that reasons over a knowledge graph can combine facts from CRM, meetings, projects and documents in one answer, and can show where each fact came from.
How does Karnil use a knowledge graph?
The Karnil knowledge graph is the brain of Karnil. It connects a company’s CRM, meetings, projects, people, documents and knowledge, and the Karnil voice assistant answers from it. Read more in What is an AI business operating system?
Updated: