Info Extraction

Industry: CMS firm

Overview

Information like name, amount, contract type, organization & many more were to be extracted from any given contract by leading global CMS’s users. It may happen that contract type for every organization may differ with differences in their structure as well. So to extract information from the uploaded contract we were tasked to implement AI & DL to work our way through with extraction with accuracy in comparison with leading cloud AI service providers.

Technology

  • Python
  • Deep Learning
  • Neural Networks
  • Transfer Learning
  • Natural Language Processing

Key Technical Challenges:

  • Working on many different kinds of documents
  • Extracting different currencies & other implemented features. In total 32 features.

Business + Technical Points:

  • Quick response from trained model
  • Scalable & adaptable model for multiple languages
  • Easy training pipeline for future language-based models.

Result:

  • We were able to extract the desired information as good as the cloud provider with lesser cost to client per API.