Metadata inference agents are designed to assist in the enrichment of metadata while simultaneously reducing the burden that documentation can be to an administrator. Agents are designed to evaluate the data and glean information from it that can be used to enhance variables metadata. Agents can be run by navigating to a specific data source, selecting the metadata inference tab, and choosing the agents desired.
When agents run they will work in one of two ways.
- The agent selected will perform specific SQL queries on the back end database to produce useful information about the data.
- If a data source does not have certain querying capabilities that the agents require there are also more basic agents that will simply run queries to retrieve the raw data from the database and then perform their own analysis on the data.
Whether the agent looks at the raw data or uses SQL queries to evaluate, all agent will produce facts. A fact is simply a finding that was produced by an agent. Facts can be evaluated by the administrators and either be disposed of, kept as a fact, or promoted to specific properties on the variable. For more information on running the agents, evaluating facts, and deleting or applying facts see: