AI Document Extraction with Context Awareness
Hero Platform_ has always made the creation of Custom Invoice Extraction models easy with our Document Studio. The difficult, and worse, time consuming portion of training an AI model with custom fields is the data preparation stage.
It takes long painstaking human hours to prepare data correctly by marking and labeling hundreds if not thousands of fields in order to generate accurate AI models that could be used to automate the extraction of your valuable data. Automation Hero has found an advanced method that can not only reduce the amount of time spent preparing data but will also increase the accuracy of the results.
The solution is our new Context Aware method using your pre-existing data to build document extraction models.
If you already have a data source storing possible values for the fields you want to extract from documents, that list of values can now be used with as few as 15 sample documents to build extremely accurate document extraction models. No more spending days, weeks, or even months with your eyeballs glued to a screen marking and labeling data. If you already have sample values stored in a data source, e.g., a database, you can use those values to drastically cut down work time as well as human error in data preparation.
Context Aware Example
This is a simple example to explain how beneficial this feature can be in building document extraction models.
You have incoming documents (PDF or images) that contain text that needs to be extracted.
The fields that need to be extracted are:
- Company names
- Invoice numbers
- Equipment names
- Total values
These are example field names. You can create any custom fields names for extraction.
Data Preparation Stage
Begin preparing your data with Hero Platform_ tools and the Human in the Loop application.
Build Your Data Extraction Model
After clicking Save and Train, match your custom fields with the Input and corresponding input field containing the sample values.
That's it. After training has been completed, use your new data extraction model in your automated Flows.