Overview

SmartSoft’s classification engine is used to automatically classify large volumes of scanned or computer-generated documents into classes.

Technology

The software uses, a subfield of machine learning called Deep Learning which models abstractions by using a deep graph, with multiple processing layers. It is also known as a neural network.

No more rules managing

This is a huge step forward compared to the existing rule-based solutions, where users need to carefully create a large number of complicated rules and periodically update them.

The Process

The software needs to be trained and is then able to classify a general range of documents into separate classes. There are several ways of training the software:

• The initial training can be done by the user as they use the system (initially most of the classifications will be low-confidence but after the first several hundred manual classifications the engine starts classifying with higher confidence)
• Training can be done by the vendor as part of the deployment
• The system can be trained for certain popular document types/document environments and include that training data as part of the software

The training data is separate from the software, so the software does not have to be redesigned in order to be used in a new document environment.

Flexible Deployment

The software can be installed in the cloud, on a hosted server, or on-premises and accessed through a web browser or a thin desktop client. Both standalone and server versions are available.

Document Classification SDK for Your Own Application

The classification engine is wrapped in a library and can be easily integrated into third-party applications.