Transformation: `transform_type`
Configure how you get data from your assets and operations
Parse
transform_type = 'parse'
The one and only transform_type
you should upload your documents into. The parse
transform will parse your PDFs, audio, HTML or any other type of data into a uniform text format, priming the rest of your document workflow.
Important: Every transformation requires at least one operation with transform_type = 'parse'
. This operation must have column_type = 'assets'.
Note: While no task description is required for parse
-type columns, you must pass in a dummy string.
Extract
transform_type = 'extraction'
Considered the most common data transformation. Used to extract information from your assets columns. You can extract data small and large, and the data can be ready-to-use (the name from a contract) or the first step of a processing pipeline (a table of transactions on an invoice).
Classify
transform_type = 'classification'
As simple as it sounds - classify data into one of several categories. Works best in pipelines downstream from extraction
columns.
Generate
transform_type = 'generation'
Used to imagine new data from existing data. Use cases including drafting short blurbs from documents - for example, a matching memo for a particular invoice.
Manual
transform_type = 'manual'
For data that you want to establish manually. Use cases include manually assigning an approver for a particular invoice.
Note: While no task description is required for manual
-type columns, you must pass in a dummy string.
API Requests
transform_type = 'api_request'
Use data anywhere along your document extraction pipeline in the body of an API request. Can only be created on the Trellis Dashboard.
Lambda Functions
transform_type = 'run_function'
Use data anywhere along your document extraction pipeline in lambda functions for deterministic, programmable results. Can only be created on the Trellis dashboard.