Event Handling with Models

In some cases, we want to have models that not only handle training and inference, but can also react to events from the platform.

An example might be a model that is used to determine which documents should be inferred in a project for example. These are not traditional models, but are really event handlers that are triggered by events in the platform. To support this, we need to do two things:

• Add a flag to our model metadata to indicate that we want to handle events
• Include a new method that will receive

Setting up a model to handle events

To set up a model to handle events, we need to add a new flag to the model metadata called eventAware and set it to true.

slug: my-model
version: 1.0.0
orgSlug: kodexa
type: store
storeType: MODEL
name: My Event Model
metadata:
  atomic: true
  state: TRAINED
  eventAware: true
  modelRuntimeRef: kodexa/base-model-runtime
  type: model

  contents:
    - model/*

Then in our model, we must add a new method called handle_event that will be called by the model runtime when an event is triggered.

def handle_event(event):
    logger.info(f"Received event {event}")

On this page

• Setting up a model to handle events