Grammar prediction - containers are now more accessible

When working with Containers creating Lexicon entries can be a big hurdle. We want to get rid of that tedious work so that you can concentrate on writing and automating great content. Containers are now provided with the necessary grammatical information without the need to cover your data values in the Lexicon.

Containers - Immensely useful, and time saving

Containers change your data values based on the grammatical function they need to fill in a statement. Examples are:

  • Automatically adding the correct determiner
  • Switching noun, verb or adjective from singular to plural
  • Changing the grammar of the verb, based on a variable noun

This reduces workload in the writer tab, because you don't need to create branches and branch triggers for different grammatical cases: Grammar is handled automatically.

You can learn more in our Container Documentationopen in new window or our Container Webinaropen in new window.

What's new

We are adding a predictive language model that generates grammatical information of your data values. This feature, called Grammar prediction, enables you to use Containers without having to prepare the necessary grammatical information.

The feature covers all aspects of a word that are grammar dependent:

  • Gender
  • Animacy (in languages where it is grammatically relevant)
  • Number
  • Inflection

Predictions are then shown in the container preview in the write tab, in the review result in the review tab and in the generated content in the results tab. That also means that the preview of your container configuration is more accurate from the get-go.

For most languages that we offer, the accuracy for both gender and providing the inflection table are close to or over 90%, meaning that if you need those grammatical features, 90% of lexicon work is covered. Most languages meaning the usual:

  • German
  • French
  • Dutch
  • Spanish
  • Finnish
  • ...

Many more are already available so feel free to test the language you want to cover next. English is not included because the regularity of the language doesn't actually need a prediction model.

How to activate the Feature

Grammar Prediction can be activated separately in each project in the project settings.

This setting is revocable in case you want to rely solely on manual lexicon entries instead.

Send us your Feedback!

Right now we are still in the process of adding new languages and improving the accuracy of our predictions. You can provide feedback by correcting wrong predictions with lexicon entries. That enables you to continue to work on the project and we would inspect the corrections and include them in our model.

We're also taking a good look at the usual workflows within your projects and will provide you with a few approaches on how you can be certain to have covered all your data, to ensure a consistent content quality.

In the meantime, we wouldn't mind if you took a good look at the current state of the feature and let us know what points are especially useful within your workflow on the platform. We would in turn inspect every lexicon entry and ensure constant update of our model to cover all cases. We're looking forward to your feedback!