A company uses a business information management system (based on the ERP/CRM concept) to manage internal processes and communicate with customers.
This product receives and allows the management of hundreds of weekly emails sent by customers and suppliers, but requires several people from the Customer Service department to fully read each and everyone of them to define the corresponding action.
The development of a new module with capabilities for NLP and machine learning automatically analyses the body of the received messages, classifies them, and launches alarms when specific needs are detected.
METHODOLOGY & results
Architecture: Cloud architecture based on Microsoft Azure, using databases such as SQL Server.
Developing language: Python, .Net (C#).
ML techniques: Supervised learning algorithms (ANN) and NLP techniques.
Results: The system allows a correct automatic classification of received emails greater than 90%, and a detection of urgent needs of around 80%.