knowledge-agents

Tractus-X Matchmaking and Inference Agent (Upcoming)

This is a folder providing a FOSS implementations of a Matchmaking Agent.

The Matchmaking Agent is needed by any Agent-Enabled dataspace participant.

The Matchmaking Agent is able to interact with and reason about the Dataspace/EDC (and hence: other Matchmaking Agents) using a Federated Data Catalogue.

It is also able to delegate work on sub-graphs/asset to tenant-internal agents (Binding Agents) providing actual data and functions.

For Agent-Enabled dataspace consumers, deploying the Matchmaking Agent alongside an Agent-Enabled EDC is sufficient.

Currently, the Matchmaking Agent (based on Apache Jena Fuseki) is baked into the EDC data plane (Agent Plane).

This will be refactored in the near future.

What is Matchmaking

A query/computation engine is federating if parts of the query maybe (recursively) dumped off to separate query/computation engines collocated in the same environment/dataspace.

Usually, there will be an explicit construct in the query language which determines (or even: infers) the required computation target. This is useful when the targets of the computations are unique in the sense that they can be listed/lookedup in advance.

But there maybe cases in which

In these cases, so-called Matchmaking agents come into play. They introduce several measures to improve the behaviour of more complex dataspaces: