Enhancing and Applying Artificial Intelligence in Industry (EnoSAGE)

The project aims to tackle the challenges of integrating conversational agent within organizations and explore innovative solutions to enhance the effectiveness.

The project at a glance

Start date:

01 May 2024
Duration in months:

36
Funding:

Enovos Luxembourg S.A. / SnT/UL
Principal Investigator(s):

Gilbert FRIDGEN

About

This industry partnership between Enovos Luxembourg S.A. and SnT/FINATRAX aims to leverage Large Language Models (LLMs) and Information Systems research to enhance customer support system. Enovos is the Luxembourg’s main energy supplier and operates in Germany, Belgium and France. Enovos invests in renewable energies and aims to continuously develop its offer in the areas of energy efficiency, decentralised power production, smart applications and mobility solutions. In this partnership, Enovos aims to leverage LLM for helpdesk ticket system for internal management. SnT/FINATRAX, with their deep expertise in Artificial Intelligence (AI) and Information Systems research, will complement Enovos SA throughout this initiative, providing valuable guidance and technical know-how. Together, they aim to drive innovation and improve efficiency in managing helpdesk operations. The collaboration focuses on three research initiatives. • Firstly, the project aims to enhance the reasoning capabilities of LLMs through knowledge graph integration. Targeting industrial requirements, the initial phase concentrates on hierarchical text classification to address challenges posed by massive, imbalanced labelled data. • Secondly, the project aims to benchmark machine learning, LLMs with knowledge graph for analyzing helpdesk tickets, detecting anomalies, and generating visual insights to support decision-making. The project aims also to develop visualizations representing patterns and anomalies. • Thirdly, the project investigates explainable and social chatbots by integrating Explainable AI (XAI) in customer support systems. The research seeks to develop chatbots capable of role-specific interactions while maintaining transparent decision-making processes through interpretable AI functionalities.