Long Titel: A Quality-Aware Explainable Ticket Resolution Assistance Framework for Industrial Customer Support

Short: 

Modern industrial environments use highly interconnected cyber-physical systems consisting of mechanical components, sensors, and complex software. In such system landscapes, service tickets are used as the central means of communication between customers and support. These tickets are traditionally processed manually by experts, which is time-consuming and costly. Although automated recommendation systems exist, there is a lack of mechanisms for evaluating ticket quality and providing transparent and comprehensible recommendations for action. This leads to low model reliability, limited user confidence, and inefficient support processes.

The Q-EXTRA (Quality-aware EXplainable Ticket Resolution Assistance) project therefore aims to develop and evaluate a data-driven framework that combines automated quality assessment of incoming tickets with explainable solution recommendations. In particular, the quality dimensions of completeness and consistency are analyzed. In addition, instructions for completing missing information are issued and transparent recommendations for action are generated with comprehensible reasoning. The confidence of the result is based on similar cases that have already been solved.

The system is tested using industrial data and evaluated in terms of accuracy and interpretability. In addition to scientific contributions at the interface of explainable AI and data quality management, the solution is designed to enable faster, more reliable, and more transparent support processes, thereby reducing downtime and costs.

Results:

The project results are structured along two work packages: WP1 – Data-Driven Ticket Quality Assessment and Enrichment and WP2 – Explainable Resolution Recommendation Framework.

WP1 will develop a model for automatic ticket quality assessment. In addition, a prototype will be implemented that predicts missing information in new tickets and provides customers with targeted feedback to improve their problem description.

In WP2, a structured classification of recommendation techniques and explainable AI methods will be created, and an architecture for the Q-EXTRA framework will be derived from this. Subsequently, an integrated proof-of-concept pipeline will be implemented that combines ticket quality assessment and explainable solution recommendations.