📝 Creating an Extraction Schema

Once you have uploaded at least one document, it's time to create an Extraction Schema. An extraction schema is structured data that describes the information you need to extract from your documents.

The extraction schema is defined at the inbox level, meaning it will be applied to all the documents within that inbox. To create a schema, open any document and select the Extraction Schema tab.

Creating an extraction schema is a simple and straightforward process. Airparser supports two types of data you can extract: simple text data (such as invoice id, dates, amounts, names, addresses, etc.) and lists/tables (like a list of ordered items, or a list of movies).

Each schema element has 3 attributes: name, description, and default value.

• The field's name is how this extracted data will be named in the parsed data.
• The description is an important field that instructs the GPT parser on what data you are looking to extract.
• The default value allows you to specify a fallback value for the field if the desired data is not present in the document.

Once you have created the schema, click the "Reprocess" button to parse your document. All the new documents you upload later will be parsed automatically.

You can reparse previously parsed documents without using additional credits if you need to trigger your integrations. However, if you update the extraction schema, Airparser needs to fully reparse the document, which consumes additional credits.