AI-assisted mapping generation

Ontopic can use AI to assist with mapping generation. When enabled, it proposes class and property mapping entries automatically, so you don't have to write them from scratch.

To see it in practice, follow the quickstart guide.

How it works

To produce relevant mapping proposals, the AI is given access to:

• The database schema
• The ontology
• Existing mapping entries
• Some sample rows from the tables

From this context, it suggests mapping entries that connect the database structure to the concepts and properties defined in the ontology. You review the proposals and decide which ones to keep, adjust, or discard.

Enabling it

AI-assisted mapping generation is disabled by default and must be enabled on a per-project basis. This gives you full control over which projects are allowed to send data to the AI and lets you leave it off for projects where it is not needed or not permitted.

To enable it:

1. In Ontopic, go to Settings > AI Configuration.
2. Click on Enable AI and select the preferred LLM model from the drop-down menu.

Server-level configuration

Before it can be enabled on a project, the AI server must be configured with access to an LLM provider. The following options are supported:

• OpenAI directly
• Anthropic directly
• An OpenAI-compatible proxy (via a custom base URL)
• A proxy protected by an OAuth client-credentials grant (token endpoint)

For concrete deployment instructions, see:

• metaphactory Kubernetes Helm charts (opens new window)
• metaphactory Docker Compose (opens new window)