Hi Tejaskumar,
The rules based model is Decipher's default training method for all documentation and is the core of its learning. This is also referred to as "Training Data".
This model will use the following 'hints' to locate information on the page:
- Keywords (e.g. sample headers)
- Data type (e.g. Date, Money)
- Format (Regex formatting)
- Lists
- Formulas
- Location (after user feedback)
Many of these hints can be defined in the DFD and will provide Decipher with a great starting position. The location data will be added as you verify documents.
The ML model uses much of the same information but will create a model specific to the document type. This will enhance the performance of the training data, but isn't mandatory for every document type.
Regards
Ben
------------------------------
Ben Lyons
Product Consultant
Blue Prism
UK
------------------------------
Ben Lyons
Principal Product Specialist - Decipher
SS&C Blue Prism
UK based