To help you get started, these tutorials provide an overview of the three main components of Fonduer. By reviewing these tutorials, you will learn how Fonduer's data model is populated and structured, how candidates are generated, and what APIs are available for you to provide weak supervision.
- Fonduer's Data Model: Learn how data is processed into Fonduer's multimodal data model, which serves as the unified representation for the rest of the pipeline.
- Generating Candidates with Matchers and Throttlers: Explore how candidates are generated in Fonduer by using matchers to define your relation types and throttlers as filtering functions.
- Weak Supervision for Richly Formatted Data: Fonduer leverages weak supervision to create large amounts of richly formatted training data. Learn the ways you can provide supervision signals in this Tutorial.
In the future, we will add additional tutorials for more advanced Fonduer features such as adding new document preprocessors and extending the data model.