IQ Bot Extraction package
Capability to automatically extract values provides enhanced content extraction from invoices.
Learning instance is an IQ Bot environment that enables you to train sample documents for content extraction. By constantly refining the learning instance, you can achieve a high content extraction accuracy before deploying the learning instance across a production environment.
IQ Bot Extraction package combines learning instance (new or existing) and a pretrained machine learning model to automatically extract content. The pretrained machine learning model uses data points to extract content from the supported document types. By providing additional training to an existing learning instance, you can extract content from other document types as well.
The following diagram shows the IQ Bot Extraction package workflow:
Contact your Customer Success Manager (CSM) or Partner Enablement Manager (PEM) for more information on the high level architecture, design and folder structure.
IQ Bot Extraction package provides the following extraction methods:
Content extraction with minimum training
Consider a scenario where you want extract sales data from various invoices. Create a learning instance by selecting a pretrained domain or document type used to extract data. In this scenario, you can select the document type as invoice, which provides a set of preset fields for content extraction. Ensure you upload a sample invoice document as a reference. Additional training of the learning instance group is not required. Use the IQ Bot Extraction package to link this learning instance to a bot. You can then run this bot to retrieve sales data based on the preset fields for various invoices.
Enhanced extraction with additional training
The IQ Bot Extraction package uses a combination of backend engine with IQ Bot server for enhanced document extraction. You can use an existing learning instance to provide additional training for all the available document groups. By customizing the various fields and validation settings of the learning instance group, you can use this package content extraction across other document types.
Before you start
- Ensure you create a learning instance using ABBYY FineReader Engine
OCR 12.4 engine.
Other OCR engines available are:
- Tesseract4 OCR
- Microsoft Azure 2.0
- Microsoft Azure 3.2
- Tegaki (not available for Automation 360 IQ Bot Cloud)
- Google Vision API
- Select documents with common layouts.
The pretrained model contains logical group. It is important to select an existing group for default validation rules to avoid documents with errors in the Success folder.
Action in the IQ Bot Extraction package
|IQ Bot Extraction||See Using IQ Bot Process documents action.|
Watch the following video to understand how to use the IQ Bot Extraction package: