Table of Content

CLIENT OVERVIEW

Our client is a multinational company providing Audit, Tax and Advisory services to organizations globally across countries. Our client works closely with organizations from diversified sectors and helps them to mitigate risk and sense opportunities. Our client also provides services in data and analytics space to help companies improve market share, anticipate tax, change strategy based on customer behaviour, and improve competitive positioning and so on.

KEY REQUIREMENTS

  • Develop an editor to upload files in different formats and view the contract
  • Develop a graphical representation of the result to compare it with the original document
  • Find relation among the entities extracted, for instance two parties involved in the contract, their address, contract amount, contract start date, contract end date and so on
  • Develop a navigation feature to showcase result entities in original contract

KEY CONTRIBUTION

  • Built rich text editor to upload and view the contract using AngularJS and jQuery
  • Designed node graph using AngularJS and jQuery to represent the entities and their relations in a much simplified manner
  • Built a customized address detection algorithm using the decision making and classification algorithm to detect addresses within the contract
  • Built provision for creating or updating domain specific and user driven increment custom model via machine learning and reinforcing framework accuracy
  • Used burly NLP techniques like phrase detection and long tail search to find the relation among the entities extracted
  • Assisted predefined contract specific text template to support context base training using machine learning algorithms

HIGH LEVEL ARCHITECTURE

TECHNOLOGY STACK

  • JavaScript
  • jQuery
  • AngularJS
  • REST API
  • Spring Framework
  • NLP Rule Engine
  • Name Entity Recognition
  • SQL Server
  • J2EE Container
  • Address Detection Algorithm

KEY BENEFITS

  • Automating the document review process improved the overall workflow efficiency by 40% resulting in quick review cycles.
  • Human errors got eliminated thereby improving the accuracy of the contract review process by 60%.
  • Feedback loop using machine learning techniques improvised the accuracy of the tool.